ECONOMIC OPENNESS AND ECONOMIC GROWTH : A COINTEGRATION ANALYSIS FOR ASEAN-5 COUNTRIES

The paper considers three channels of economic openness, namely FDI, imports, and exports, and examines their short-run and long-run effects on the economic growth in the five founding member countries of the Association of Southeast Asian Nations (ASEAN) over the period from 1980 to 2014. Besides the impact on the economic growth, the authors analyze all possible causal interrelationships to discern patterns and directions of causality among FDI, imports, exports, and GDP. The quantitative analysis, which is based on the vector error correction co-integration framework, is conducted separately for each country in order to assess their individual experiences and allow for a comparative view. Although the precise details differ across countries, the findings indicate that there is a long-run equilibrium relationship between economic openness and GDP in all ASEAN-5 economies. FDI, imports and exports have a significantly positive short-run and long-run impact on the economic growth. Our results also show that export-led growth is the most important economic growth factor in most countries, followed by FDI-led growth. Another crucial finding is the bi-directional causality between exports and FDI across the ASEAN-5 countries. This indicates the presence of direct and indirect effects on GDP and a self-reinforcing process of causality between those two variables, which strengthens their impact on the economic growth.


INTRODUCTION
From the theoretical as well as empirical point of view, it has been long suggested that international economic integration promotes effi cient reallocation of resources leading to the increased economic growth and sustainable development.Economic openness in developing countries can provide numerous opportunities for attracting foreign capital and undergoing structural transformations, which is vital for economic modernization and sustainable growth.
Notably, the member countries of the Association of Southeast Asian Nations (ASEAN) are regarded as particularly open economies, embracing international integration as a national development strategy.ASEAN, which has undergone several steps of regional integration since 1967, is indeed a very dynamic region with regard to intra-regional and global economic integration.More specifi cally, the fi ve founding ASEAN members have been subject to integration over a long period.International trade as well as foreign direct investments (FDI) play a signifi cant role in the national economies of those countries.
Consequently, it may be argued that the experiences of the ASEAN countries off er empirical evidence and insights into the eff ects of various channels of economic openness and growth as well as on all relevant interrelationships.As FDI, imports, and exports represent the most important dimensions of international economic integration, information on those aspects of openness is particularly important.Besides country and regionalspecifi c information, the ASEAN case might also provide relevant lessons and policy implications for other developing and emerging economies opening up and following a similar path.In an increasingly globalized world economy, knowledge on the long-run links and eff ects among diff erent dimensions of economic openness and economic growth is valuable for economic policy purposes.
Th us, our paper primarily aims to trace the long-run trends in FDI, international trade, and economic growth in fi ve founding ASEAN member countries, for which a comprehensive dataset can be constructed over the period 1980-2014.Secondly and most importantly, it aims to identify, analyse, and assess relevant long-run and shortrun causal economic relationships among FDI, imports, exports, and GDP.Our quantitative analysis pays particular attention to the economic growth eff ects of each of the aforementioned channels of economic openness.
Th e rest of the paper is structured as follows.Th e next section provides an overview of relevant theoretical and empirical literature.Th e third section discusses the data and examines the long-run trends.Th e fourth section presents the methodology and modelling framework of the quantitative analysis.Th e fi ft h section reports and discusses the empirical results.Th e fi nal section summarizes the key fi ndings and provides concluding remarks.

THEORETICAL AND EMPIRICAL BACKGROUND
Long-standing and extensive theoretical literature on international trade advocates that trade contributes positively to the long-term growth prospects of a country (e.g.Van den Berg & Lewer, 2015).Integration and cross-border trade promote economic effi ciency through resource reallocation, increased productivity, production specialization, scale economies and increased market opportunities.
Th ere is also accumulated empirical evidence suggesting that international trade has led to the increased effi ciency, productivity, technology, and growth in both developed and developing countries (Th e World Bank, 2015;Feenstra, 2015;UNCTAD, 2015;Markusen et al., 1995).
Similarly, the literature considers inward FDI to be particularly important for developing countries as it provides productive capital necessary for performing economic activities (e.g.UNCTAD, 2016).Th us, it adds to the domestic capital stock and enhances the production capabilities of the national economy in the long-run.Moreover, it is suggested that FDI may lead to the transfer of technology, economic transformation and technological progress.Other possible positive spillovers include the creation of linkages between multinational enterprises (MNEs) and local fi rms, resulting in increased domestic entrepreneurship.
Lastly, besides the above channels, FDI can also contribute to economic growth through increased exports, if MNEs locate in the host-country for export activity.Th ere are many empirical studies investigating the long-run relationships among economic openness, trade and growth.Generally, those studies, which are usually based on a time-series framework and co-integration analysis, fi nd evidence of a signifi cant growth eff ect of international integration related factors.
For instance, Gries and Redlin (2012), using a large sample of 158 countries over the period 1970-2009, conclude that there is a signifi cant positive causal long-term eff ect running from trade openness to economic growth and vice versa.Using various trade liberalization indices for 120 countries over the period 1970-1999, Yanikkaya (2003) ) fi nds that trade has a positive impact on real GDP per capita.
For Southeast Asian economies, where trade liberalization has been on the rise for many decades, international trade has been a contributing factor of economic growth and development.Several studies have found empirical evidence of a long-run link between exports and GDP (Ridzuan et al., 2016;Tan & Tang, 2016;Hsiao & Hsiao, 2006;Ahmad & Harnhirun, 1996).However, depending on the given countries and time period concerned, the evidence sometimes gives clear support to the export-led growth hypothesis.In other instances, it is revealed that there is a more complex structure indicating GDP causing exports or the existence of a bi-directional causal relationship.
Besides trade openness, there are studies that also examine explicitly the long run causal relationship between FDI and growth.More specifically, the empirical evidence points out that in Southeast Asian countries FDI contributes significantly to GDP growth (Bhatt, 2014;Moudatsou & Kyrkilis, 2011;Choong & Liew, 2009).Additionally, studies in which trade-related measures and FDI are included simultaneously to investigate long-run growth eff ects, reveal the presence of intricate interrelationships (Sothan, 2016;Tan & Tang, 2016;Bhatt, 2014).It is found that diff erent patterns and causality relations can arise.
For instance, FDI might cause trade (exports and imports), which causes economic growth without a direct growth eff ect of FDI, or FDI and exports both having only direct eff ects on growth, or other interdependencies.Th e literature shows that as international integration issues and growth can be interrelated in various ways, the establishment of relevant patterns and long-run relationships is essentially country specifi c.Th is also represents additional motivation for our empirical modelling approach on the underlying interdependent relationships and causalities to be based on an individual country case investigation for all ASEAN member states in our dataset.

DATA, VARIABLES AND LONG-RUN TRENDS
For each of the 5 ASEAN countries (Indonesia, Malaysia, the Philippines, Singapore, and Th ailand), the data cover the period 1980-2014.Th ere are four key variables expressed in US dollars: All variables are expressed in real terms, by defl ating the nominal values using the country's GDP defl ator.Th e base year is 2005.Th e data sources are as follows: GDP from UN national accounts database; FDI stock from UN FDI database; exports, imports, and GDP defl ators from the IMF international fi nancial statistics database.
Prior to our main analysis, we proceed with a brief descriptive analysis and discussion on the long-term trends of the observed variables for each ASEAN-5 country.As evident from Figure 1, there is clearly a strong upward trend in real GDP, real exports, real imports, and real FDI over the period 1980-2014 in all of the ASEAN member states in our dataset.Macroeconomic variables related to economic openness increase along with GDP growth.
In terms of value, real exports and real imports are the most important international integration related measures, followed by real FDI.Th e exception is Singapore where foreign direct investment has become more important and has surpassed exports and imports in the last nine years of the observed period.In summary, we can see that economic openness is important and is rapidly increasing over time in ASEAN-5 countries.

METHODOLOGY AND MODELLING FRAMEWORK
As shown in the descriptive analysis, there are signifi cant long-term trends in the three interna-tional openness related measures as well as in economic growth.Most importantly, though, based on the theoretical grounds, certain long-run relationships among those variables should exist.In general, as already discussed and suggested by the relevant literature, exports, imports, and FDI, play an important role in encouraging and contributing to the economic growth.However, the importance, strength, and direction of the causal relationships among those variables may vary signifi cantly in diff erent economies.
Essentially, the interrelationships are dependent on the underlying structural relationships and dynamics of a given economy.For instance, Figure 1.Long-run trends, 1980Long-run trends, -2014 in some countries exports might be particularly infl uential with a clear unidirectional eff ect on GDP, suggesting an export-led growth hypothesis.Similarly, there might be a strong FDI-led growth scenario.In other cases, more intricate interrelationships might be at work, such as, direct and indirect eff ects among exports, imports, and FDI on growth, or bi-directional eff ects between openness and GDP growth.Th us, the underlying long-run relationships have to be discerned and established through the appropriate empirical time-series modelling.
Th e paper applies a three-step methodological approach within a vector error correction co-integration framework to analyse the causal relationships and eff ects among real GDP, real FDI, real imports and real exports in the observed ASEAN countries.Th is analytical framework is particularly appropriate when investigating the issue of long-run causal equilibrium relationship(s) among several time-series variables.It allows the effi cient estimation of long-run and short-run eff ects.It also takes explicitly into account and analyses the short-run adjustments to the empirically established long-run equilibrium relationship that exists among the covariates.
First, we examine the statistical stationary properties of the data set by applying the Augmented Dickey Fuller (Dickey & Fuller, 1981), ADF, and the Phillips and Perron (1988), PP, unit root tests.
It has to be noted that a non-stationary time series data set produces spurious results.Th erefore, stationarity is an important issue.If the data are not stationary, we take the fi rst diff erences and test again for stationary properties.In case the fi rstdiff erenced data are stationary, we can proceed with our co-integration and Granger causality analyses.
In the second step, we test for co-integration and in case the variables are found to be co-integrated, estimate the long-run equilibrium relationship.Notably, if the time series covariates under study are not co-integrated, an equilibrium relationship cannot be assumed and Granger causality tests are spurious (Granger, 1988).When two time series are I(1), integrated of order one [stationary fi rst-diff erenced series], and there exists a linear combination between those series that is I(0) [stationary series], then those two time series variables are co-integrated (Engle & Granger, 1987).
In the third stage, we analyse the short-run (Granger) causalities among all of the four variables.Th e co-integrating equilibrium relationship among GDP, FDI, IMP, and EXP, as well as the Granger causality tests, are based on the appropriate vector error correction model (VECM) specifi ed and estimated for each of the fi ve ASEAN countries.
Our VECM for a single country can be represented as follows: EJAE 2016  13 (2)  10-20 Vogiatzoglou, K., Nguyen, P. N. T.  Economic openness and economic growth where t=1,2,…T refers to the time periods (annual observations); ∆ denotes the fi rst diff erence operator; q is the number of lags; the δ's are the parameters to be estimated; ε t-1 is the error correction term, indicating the long-run causal eff ect (of the right-hand side variables on the left -hand side variable) and the speed of adjustment towards the long-run equilibrium, which is derived from the co-integration relationship; u is the serially uncorrected error terms; i, j, k and l denote the appropriate lag lengths for the VECM, based on Akaike's information criterion.
In the real GDP equation (1a), the Granger short-run causalities from real FDI, real imports, and real exports to real GDP are tested.Th e relevant three null hypotheses in (1a) can be expressed compactly as: (All short-run parameters are equal to zero) Each of the above represents Wald tests of the null joint hypothesis that the estimated short-run coeffi cients associated with either FDI, IMP, or EXP are (not statistically diff erent from) zero, and hence there is no Granger short-run causality between those variables and GDP.
Similarly, in the real FDI equation (1b), the tests for short-run causality running from GDP, IMP, and EXP to FDI are based on the following joint hypotheses: If a joint null hypothesis cannot be rejected, it is concluded that the variable associated with the given Wald test (here GDP, IMP, or EXP) does not Granger cause FDI.
In the real imports equation (1c), Granger shortrun causality tests are based on the following hypotheses: Finally, with respect to real exports, equation (1d), the relevant tests for Granger causality are represented as follows:

EMPIRICAL RESULTS
Table 1 reports the results of the unit root tests for four variables in fi ve observed ASEAN countries.Th e ADF and PP tests indicate that the four series are I(1) processes at the 5% signifi cance level except for the data about FDI in the Philippines but in the PP test, the data is I(1) at 1% level.Th us, we accept the results of the PP test for FDI variable in the Philippines.Th erefore, we may reasonably conclude that all variables appear to be I(1) series in all of the fi ve countries.
Next, using the Johansen's maximum likelihood approach, a co-integration analysis is conducted to investigate whether there is a long-run relationship among real GDP, real FDI, real exports, and real imports.Table 2 presents the results of the co-integration analysis.We fi nd evidence of co-integration in all fi ve countries, as revealed by both the maximum eigenvalue and trace statistics at the 5% and 1% statistical signifi cance levels.Th ese fi ndings support the view that GDP, FDI, IMP, and EXP all move together in the long run in each of the fi ve ASEAN countries observed herein.
Table 3 reports the results of the co-integrating regression equation, with GDP as the dependent variable.Th ese fi ndings indicate the underlying equilibrium relationship and long-run eff ects of FDI, EXP, and IMP on GDP.
It is evident that in all countries the three variables, which are associated with diff erent channels of economic openness, have a statistically significant long-run impact on GDP in all countries.Since all variables are measured in the same units (namely US dollars), the estimated regression coeffi cients reveal the magnitude of the eff ect, as well as relative importance of each variable.Exports, followed by inward FDI, are found to exert the Note: *** and ** denote statistical signifi cance at the 1% and 5% levels, respectively.

Table 2. Results for co-integrati on tests
strongest eff ect in several countries.More specifically, in the Philippines, Singapore, and Th ailand, exports exhibit the highest coeffi cient, and FDI the second highest in two of those countries.In Malaysia, FDI is associated with the highest im-pact on GDP followed by imports.Th ese fi ndings give support to the export-led and FDI-led growth hypothesis.Interestingly, imports are found to be the most important international integration related factor of economic growth in Indonesia.ECT denotes error correction term, and indicates the long-run causal eff ect (long-run equilibrium relationship) between the "explanatory" and the "dependent" variables in each equation.***, **, and * denote statistical signifi cance at the 1%, 5%, and 10% levels, respectively.

Table 4. Results for Granger causality tests
Additionally, in this country exports show the second highest impact, whilst FDI has the smallest long-run eff ect on GDP.
Having established and examined the long-run equilibrium relationship, we turn to short-run ef-fects and Granger causality tests among all considered variables, as shown in Table 4. First, it is found that FDI, imports, and exports signifi cantly Granger cause GDP in the short-run [equation ( 1a)] in all ASEAN-5 members.An exception is

Table 3. Long-run equilibrium relati onship between GDP, FDI, Imports, and Exports
Indonesia and the Philippines, where exports in the former and imports in the latter do not have a short-run impact.As the error correction term is highly statistically signifi cant, it is confi rmed that those variables cause GDP in the long-run across all countries in our sample.
By examining all causality directions among the variables (across all equations of the system, VECM), certain interesting patterns of short-run causality interrelationships can be observed.FDI is Granger caused by GDP (in Malaysia, the Philippines, and Th ailand), by imports (in Singapore and Th ailand), and exports (in all ASE-AN-5).With respect to imports, there is short-run causality from GDP (in Malaysia and Singapore), from FDI (in Indonesia, Malaysia, Singapore and Th ailand), and from exports (in all ASEAN-5).Exports are Granger caused by GDP (in Singapore), by imports (in all ASEAN-5), and by FDI (in all ASEAN-5).Th us, besides the impact of the economic openness variables to GDP, there is the evidence that in certain countries GDP causes some of those variables.However, this result is rather case specifi c and less generalizable.A much more universal pattern across the ASEAN-5 is the bidirectional causality between exports and imports and between FDI and exports.
Coupled with the fi nding that FDI and exports cause GDP in the long run and short-run, the bidirectional causality between FDI and exports implies that those two variables have a direct as well as indirect eff ect associated with their long-run impact on GDP.Th is is due to their close interrelationship, suggesting that FDI and exports cause growth directly and indirectly through their mutual impact which reinforces their eff ect on GDP.

CONCLUSIONS
By applying the vector error correction cointegration framework for fi ve ASEAN countries over the period 1980-2014, this study observes the causal relationships among FDI, imports, exports and GDP.An underlying long-run equilibrium relationship among the three aspects of economic openness and growth is found in all of the observed countries.In general, FDI, imports and exports signifi cantly aff ect and lead to the economic growth both in the short and the long-run.
However, our analysis indicates that there are some substantial diff erences across countries with respect to both the relative importance of the longrun eff ects on GDP and the patterns of short-run causalities.Th is fi nding justifi es our modelling approach of conducting a co-integration analysis for each ASEAN country separately (instead of panel co-integration).Th is highlights that international integration aff ects diff erent ASEAN countries to some extent through diff erent channels of economic openness.
Th ough certain dissimilarities are present, the overall pattern still emerges.More specifi cally, both export-led growth and FDI-led growth are found to be the two most important growth factors across the ASEAN-5.Furthermore, there is a strong interrelationship between imports and exports, suggesting that a country's overall trade openness has a favourable impact on exports.
Most importantly, a bi-directional causality between FDI and exports is evidenced in all ASE-AN-5 members.It points to the existence of direct and indirect eff ects on GDP associated with those variables.Economic growth is, thus, caused via a direct path by exports (or FDI) as well as through an indirect path, in which exports cause FDI (or FDI causes exports), which in turn cause GDP.Moreover, because of this circular (bi-directional) self-reinforcing causal process among FDI and exports, their growth impact is strengthened.Attracting inward FDI will lead to GDP growth and increased exporting activity, which shall result in GDP growth and increased inward FDI.
Th is fi nding might also suggest a scenario whereby the MNEs have to a large extent an export-platform FDI motive and are attracted to ASEAN countries due to their trade openness and export-oriented economies.In such a case, a country achieving an increasing export activity becomes increasingly the target of export-oriented MNEs, which contributes to a country's further openness.Hence, on their own as well as through their bi-directional interrelationship and interaction, FDI and exports are found to play a particularly important role as factors of economic growth in ASEAN countries.

Table 1 .
Results for unit roots tests Note: ***, **, and * indicate that the null of the unit root in the ADF and PP tests is rejected at the 1%, 5%, and 10% levels, respectively.Lag lengths are chosen using Akaike's information criterion.