TRADE OPENNESS AND ECONOMIC GROWTH: A PANEL DATA ANALYSIS OF BALTIC COUNTRIES

Melina Dritsaki¹⁺ --- Chaido Dritsaki²

¹ University of Oxford, Oxford, UK.
² Department of Accounting and Finance. University of Western Macedonia, Kozani, Greece.

Keywords:Trade openness, Economic growth, Panel causality test, Baltic countries.

JEL Classification: C23, C33, F10, F43.

ARTICLE HISTORY: Received: 27 November 2019, Revised: 14 January 2020, Accepted: 17 February 2020, Published: 24 March 2020

Contribution/ Originality: This study contributes in the existing literature in the causal relationship between trade openness and economic growth. This study uses new estimation methodology of the second generation unit root test by Pesaran (2007) non-causal Granger test for heterogenous panel data by Dumitrescu and Hurlin (2012).

1. INTRODUCTION

The past few years, trade openness plays an increasingly significant role in economic growth. The traditional models developed by Smith and Ricardo, explain that trade openness should always be followed by policies that expand the provision of public education, vocational training, health care and infrastructure. Moreover, they supported that an extensive policy change towards trade openness should be gradually reinforced, in order for people to adjust to the increased levels of international competition and technological progress. Hence, countries specialising in goods and services production have comparative advantages when it comes to exporting goods and services. On the other hand, countries which don’t have these advantages, will import from these countries and will specialise in other forms of goods and services. That leads to the resources distributions in the optimum way. According to the Heckscher-Ohlin model, a country, which exports goods and services, is using these resources intensively. As a result, trade openness will increase for this country and the economy will swift into respective sectors to sustain the increase in productivity (see López (2005)).

The neoclassic productivity model by Solow (1956) and (Ramsey, 1928) shows that the continuing economic growth is a result of an exogenous production factor, that is, the passage of time. The neoclassic production function used in this theory, is related with the exports, capital equipment, as well as labour. If new technologies manage to increase labour productivity and capital and prevent a decrease in the rate of return on investment, then labour force will grow at an exogenous rate. This part of output growth which cannot be explained by the production coefficients is usually know as Solow residual or as the total productivity of the applied work factor. In conclusion, one would claim that neoclassic models suggest that technological progress and the stable development rate of exports are completely exogenous.

The theory of endogenous development suggests that economic development is primarily due to endogenous and not exogenous forces. In other words, we would say that the theory implies that investing in human capital, technology and innovation contributes significantly to economic development. Moreover, endogenous theory suggests that the long-run development rate depends on the economic policy measures every government takes. For example, funding for research and development increases the motivation for innovation and as a consequence the rate of development. In the model Romer (1986) and Lucas (1988) developed, they support that technology includes both human and natural capital. Moreover, they support that in that case, diminishing returns to capital do not exist. Grossman and Helpman (1991) develop a model envisioning both technology and foreign trade engagement in an endogenous manner. Also, Levine and Renelt (1992) showed that trade openness encourages direct foreign investments leading to the increase of long-run growth. Developed countries, develop their productivity by utilizing new developments. Also, based on the endogenous development theory, an increase in market openness, improves technology and increases productivity.

The relationship between trade openness and economic growth has been investigated for many years. That is because trade openness, which is represented by the ratio of imports, exports, or imports plus exports to the GDP, is the motivating force behind growth in developing countries. Henceforth, the direction of the causal relationship between trade openness and economic growth is worth being investigated.

The causal relationship between trade openness and economic growth means that the growth and industrialisation in these countries have been achieved in the form of external education together with increase in trade. In this case, the export-led growth hypothesis in the neoclassical approach is advocated to be valid. This outcome will increase productivity through exports by increasing the economies of scale. Moreover, new investments are taking place and therefore an increase in employment and real wages is observed. Nowadays, trade openness is considered to have a positive influence on growth.

The structure of the paper is as follows: Section 2 briefly reviews the literature. Section 3 presents data and variables. This is followed by section 4 which presents methodology. Empirical results are discussed in section 5. Concluding remarks are given in the final section.

2. LITERATURE REVIEW

The empirical findings of the relationship between trade openness and economic growth are a debatable matter in the international economic literature. A number of studies were conducted in various countries in order to investigate the causal relationship between trade openness and economic growth. Very few of them, though, have investigated this relationship using cross-sectional data. Such is the study by Dar and Amirkhalkhali (2003) which analysed the relationship between trade openness and economic growth in 19 OECD countries during the period 1971-1999. The growth accounting model applied was estimated with the random coefficients approach using time series and cross-sectional data. The study results showed that the impact of trade openness in production growth and consequently in economic growth, differs between countries.

Gries and Redlin (2012) study a total of 158 countries and questioned the causal relationship between per capita GDP growth and trade openness during the period 1970-2009. In their study, they used panel cointegration tests and panel error-correction models (ECM) which were estimated using the GMM method of testing the causal relationship between economic growth and trade openness. The long-term outcomes suggest a positive one-way causality from trade openness towards economic growth. However, the short-term coefficients identified a negative short-run adjustment.

Zeren and Ari (2013) analyse the relationship between trade openness and economic growth in the case of the G7 countries; Germany, France, Canada, Japan, Italy, the United States, and the United Kingdom for the period 1970-2011. To test for causal effects between variables, the panel causality test introduced by Dumitrescu and Hurlin (2012) was used for heterogeneous panel data models with fixed coefficients. The study results conclude that there is a bidirectional causal relationship trade openness and economic growth.

In their study, Muhammad and Jian (2016) investigated the relationship between trade openness and growth in selected Muslim countries. The choice of Muslim countries has been down in order to investigate economies with different religious, economic and social characteristics. The study used a random and fixed effect model as well as cointegration tests by Pedroni and Kao for the long-term relationship between the variables under examination. The results from random and fixed models show that trade openness has significant and positive effect on economic development for all Muslin countries under consideration.

Mangir, Kabaklarlı, and Ayhan (2017) analysed data from 10 African countries during the period 1990-2015. Using the pool mean group estimator (PMG) and the panel Autoregressive Distributed Lag (ARDL) model, they conclude that trade openness has a positive impact on economic growth in the long run for all the countries under investigation.

Finally, Alam and Sumon (2020) studied the causal relationship between economic growth and trade openness in the case of 16 Asian countries between 1990-2017. They applied panel cointegration and causality approaches in order to examine the causal relationship between the underlined variables. The panel vector error correction model Granger causality shows a bidirectional causal relationship between economic growth and trade openness.

Dritsaki (2015) examined the impact of trade openness and foreign direct investment on economic growth for three Baltic countries during the period 1993-2011. The impact of variables on economic growth is achieved using three models; pooled Model, Fixed Effects Model (FEM) and Random Effects Model (REM). The results of the paper present an important effect of trade openness and foreign direct investment on economic growth, while Random Effects Model, which is selected for conducting the panel data analysis, denotes that trade openness has an impact on economic growth for the three Baltic countries.   

Silajdzic and Mehic (2018) investigated the impact of trade openness on economic growth in Central and Eastern European (CEE) countries for the period 1995-2013. For the analysis on their paper, they use two different estimation methods. Fixed effect panel which is applied by estimating Prais-Winsten-correlated panels corrected standard errors (PSCE) method due to the presence of autocorrelation and heteroscedasticity and the dynamic least squares dummy variable method (LSDVC). The results of their paper showed that trade openness not only from exports side but also from the increase of imports from technologically innovative EU countries, is positively connected with economic growth.

(Guei & Roux, 2019) examined the relationship between trade openness and GDP per capita among 15 countries of Economic Community of  Western  African  States  (ECOWAS)  for the period 1990-2016. Their initial analysis was based on a regression model for each of 15 examined countries which tried to find a long run relationship between variables. Afterwards, they investigated the relationship between trade openness and GDP per capita using Pooled Mean Group (PMG) models and the Autoregressive distributed  lag  (ARDL) model. Their results showed a long-run relationship between variables except for Ghana, Guinea-Bissau, Mali, Senegal and Togo. Furthermore, estimations indicate that trade openness has negative effect on GDP per capita in the long run.  

Bonga-Bonga and Kinfack (2019) assessed the relationship between trade openness and economic growth for 38 countries of Africa using annual data for the period 1970-2016. On their analysis, they employed a Panel Smooth Transition Regression (PSTR) model which represents non linearity and endogeneity on the related variables. The results present that African countries are not homogenous mainly on trade openness and economic growth. Also, the paper denotes that the relationship between trade openness and economic growth varies from country to country according to their development rate. Conversely, for middle and high income countries, there is a positive relationship.

3. DATA

This study, we investigated the causal relationship between trade openness and economic growth in the three Baltic countries (Estonia, Latvia, Lithuania) for the period 1990–2020. Economic growth (EG) is measured using per capita GDP with constant 2015 EUR, and trade openness (TO) is measured exports plus imports as a share of GDP. The data used in the paper are sourced from the AMECO. Both variables are employed with their natural logarithms.

EXP=Exports goods and services at 2015 prices.
IMP=Imports goods and services at 2015 prices.
GDP=Gross domestic product at 2015 reference levels.
GDPH=Gross domestic product at 2015 reference levels per head of population.

4. ECONOMETRIC METHODOLOGY

4.1. Dumitrescu and Hurlin Causality Test

The current study uses Dumitrescu and Hurlin (2012) causality test, which is a new test of non-causality of Granger (1969) for models with heterogenous panel data and constant factors. This test takes into account two dimensions of heterogeneity.

• The model heterogeneity of the regression used for the Granger causality test.
• The heterogeneity of the causality relationships.

The alternative hypothesis is defined as the Heterogeneous Non-Causality hypothesis-(HENC)]. In the alternative hypothesis we have two cross-sectional sub-categories.

There is a causal relationship from variable x towards variable y for the first sub-category, but it is not based necessarily in the same regression model.

For the second sub-category, there is no causal relationship from variable x towards variable y. We assume a heterogenous category of panel data in a model with constant coefficients  (during the whole period)in this category. This alternative hypothesis is the following:

Also, the standardized test statistic for fixed Tsamples is as follows:

4.2. Cross Sectional Dependence Test

The causality test by Dumitrescu and Hurlin (2012) assumes that stationarity of variables y and x. Before applying unit root tests in panel data, we need to examine the existence of a dependence between the cross-sectional unit and the variables of the model under examination. The cross-sectional non-dependence argues that the error terms are not cross-correlated and the zero error covariance is a very significant factor in the unit root tests of the panel data and the cointegration tests. For the cross-sectional dependency tests between the units we use tests by Breusch and Pagan (1980) LM, Pesaran scaled LMs (Pesaran., 2004) Pesaran CDp, and Baltagi, Feng, and Kao (2012) bias-corrected scaled LMBC. In all above tests the null hypothesis could be define as:
Η0: There is no cross-sectional dependence between the units.

4.2.1. Breusch-PaganLM test

The most well-known diagnostic test for cross-sectional dependency is the test by Breusch and Pagan (1980) using the Lagrange Multiplier (LM) test statistic.

4.2.2. Pesaran LMs  ScaleTest

In the case of many cross-sectional units N, Breusch-Pagan test is not reliable. Hence, Pesaran. (2004) suggested a standardized version of Lagrange Multiplier (LM) which will take the form below:

4.2.3. PesaranCDpTest

In order to deal with the differentiation of the magnitude of cross-sectional units with the Lagrange LM and LMS statistics, Pesaran. (2004) suggested an alternative statistic, based on the average of the pairwise correlation coefficients given by:

4.2.4.  Baltagi, Feng, and Kao LMBC  Test

Baltagi et al. (2012) recommend an amendment into Pesaran. (2004) for the scale of Lagrange Multiplier test (LMS), which take the form below:

Once the dependence or non-dependence between the cross sectional units is determined, we move on to the first generation tests (non-dependence) or the second generation tests (dependence) of cross-sectional units.

4.3. Panel Unit Root Test

4.3.1. First Generation Panel Unit Root Tests (Cross-Country Independence)

In the first generation category, we test for the homogeneity hypothesis with Levin, Lin, and Chu (2002) test, while the heterogeneity hypothesis is tested with Im, Pesaran, and Shin (2003) test.

4.3.1.1. Levin-Lin-Chu test

Assume a variable, which is detected in N countries during T periods, and also assume a model with individual effects. Levin, Lin and Chu test examines a model whose lagged dependent variable is homogenous across all panel units.

4.3.1.2. Im, Pesaran and Shin

Im et al. (2003) test is based on the cross-sectional independence assumption. In contrast to  Levin et al. (2002) test, Im et al. (2003)  test allows the hetegeneity in the value of i  under the alternative hypothesis. Im et al. (2003)  use Levin et al. (2002) model (6) and replace p with p_i. Hence, their model with individual effects could be written as:

If this statistics is correctly standardised, then it asymptotically follows a normal distribution.

4.3.2. Second Generation Panel Unit Root Tests (Cross-Country Dependence)

If a cross-sectional dependency exists in panel data, then the first generation unit root tests cannot be used. In this case, we use the second generation unit root tests, such as SURADF, CADF and CIPS tests.

4.3.2.1. The Pesaran Tests

In the case that a cross-sectional dependence, Pesaran (2007) suggests a one-factor model with heterogeneous loading factors for residuals for the unit root test. This model instead of relying on unit root tests on deviations from the estimated common factors, increases Dickey-Fuller (ADF) model with the cross section average of lagged in the levels and the first difference of the individual series.

If the residuals are not serially correlated, then the regression used for the ith unit is given by:

5. EMPIRICAL FINDINGS

5.1. Dumitrescu and Hurlin Causality Test

In order to determine the causal relationship between trade openness and economic growth, according to Dumitrescu and Hurlin (2012) test, coefficients on panel data should be stationary. Prior to applying the unit root test on panel data, we examine if a dependence relationship exists among the cross-sectional units in the model variables under examination. Table 1, presents the results of the cross-section dependence test for the variables.

Table-1. The results of cross-section dependence test for the variables.

Variab.	LM		LMs		CDp		LMBC
	Stat.	Prob.	Stat.	Prob.	Stat.	Prob.	Stat.	Prob.
LGDPH	86.68	0.000	34.16	0.000	34.11	0.000	9.38	0.000
LTO	86.46	0.000	34.07	0.000	34.02	0.000	9.29	0.000

Note: The null hypotheses of tests are of presence of no cross sectional dependence in panel.

Results of the table above show that there exists a cross-section dependence across countries in the 1% significance level for the time series LGDPH (Gross Domestic Product per head of population) and LTO (Trade Openness).

Therefore, the second-generation unit root tests should be applied. On the basis, that cross-sectional dependence between countries has been defined, the second generation of panel unit root tests (Pesaran’s CIPS test) should be applied. This procedure begins with the OLS estimation for the ith cross section in the panel considering the following Cross-Sectional Augmented Dickey Fuller (CADF) regression. Table 2 presents the results of the second generation of panel unit root tests CIPS.

Table-2. Second generation of panel unit root tests (pesaran’s cips test).

Variables	LGDPH	LTO	Critical values
			1%	5%	10%
Intercept	-3.12	-2.97	-2.60	-2.34	-2.21
Intercept and Trend	-4.64	-6.15	-3.15	-2.88	-2.74

Note: CIPS: Cross-sectionally augmented IPS.
Critical values have been obtained from (Pesaran, 2007).

According to the Pesaran CIPS test result presented in Table 2, the null hypothesis of all panels contain unit roots is rejected at the 1%, 5% and 10% significance level. Therefore, we could argue that the variables are stationary and as a result, we could define the causal relationship between trade openness and economic growth based on the Dumitrescu and Hurlin test. The Dumitrescu and Hurlin causality test results are presents in Table 3 below.

Table-3. The results of dumitrescu and hurlin panel causality test.

6. CONCLUSION

The relationship between trade openness and economic growth is considered a controversial one in the economic literature. It has received greater attention particularly in the past few years, considering the difference in economic performance especially between the developing countries. The difference in the catching-up processes between transition economies, such as those of Baltic countries (Estonia, Latvia, Lithuania), the form and their dynamics of joining into the European and international economic structures, remained to a great extent difficult to explain due to the unexplained different economic, political and institutional factors that followed. This issue becomes even more controversial, on the basis that transitional economies have followed similar process towards economic liberation during the first years of their transition period. Henceforth, the benefits of trade openness remain controversial and being discussed even further in the international and academic political debates.

Nonetheless, trade openness is considered a significant and determinant factor for wages and economic growth. Trade integration allows the more effective allocation of resources through economies of scale as well as through increased competition. It facilitates knowledge diffusion and technology transfer, which affect the cost and productivity standards, which promote technological advancement and lead to greater efficiency. In this framework, it is generally believed that as trade openness  increases, economic growth  is expected to increase with the rise of technology innovation and productivity.

The current study investigated the causal relationship between market openness and economic growth in the case of Baltic countries. For that purpose we employed the non-causality Granger test for heterogenous panel data, which was developed by Dumitrescu and Hurlin (2012). The findings of the study confirm the existence of a cross-sectional dependence of model time series between the countries under examination, which confirms that Baltic countries share common factors and common ties. The time series dependence led us to use the second-generation unit root test by Pesaran (2007) which showed that time series are stationary. Therefore, we could use the Dumitrescu and Hurlin (2012) test which meets both prerequisites; cross-sectional dependence and stationarity. The results of Dumitrescu and Hurlin test revealed a bidirectional causality between trade openness and economic growth in Baltic countries. The use of Dumitrescu and Hurlin test contributes to this study being differentiated from previous ones which investigated the causal relationship between trade openness and economic growth.  Baltic countries experienced a strong economic growth and a rapid closure of income gap with developed countries until the beginning of global financial crisis. Since then, Baltic countries have experienced a deceleration of economic growth. The later led to the discussion of whether Baltic countries could get away from the trap of mediocre growth. Mediocre growth could be due to lack of coordination between various economic factors which hold back production and market openness. Such result could be alarming as unfavourable prospects could reduce investments in natural capital, human capital and organisation progress.

Those responsible for engraving policies could assist in easing production for a quicker and more stable growth, taking measures to tackle structural coordination issues in Baltic countries. The production that relies on knowledge requires investment in humans and human capital. Also, it needs reforms which reinforce education and accumulation of abilities in every level. Finally, investments in capital equipment and intellectual capital will primarily burden private businesses so the availability of funding is essential. The current study focused on Baltic countries, but in essence the discussions and methodology of this study could be applicable and transferable to other countries, such as the countries of Central and Eastern Europe.

Funding: This study received no specific financial support.

Competing Interests: The authors declare that they have no competing interests.

Acknowledgement: Both authors contributed equally to the conception and design of the study.

REFERENCES

Alam, K. J., & Sumon, K. K. (2020). Causal relationship between trade openness and economic growth: A panel data analysis of Asian countries. International Journal of Economics and Financial Issues, 10(1), 118-126. Available at: https://doi.org/10.32479/ijefi.8657.

Baltagi, B. H., Feng, Q., & Kao, C. (2012). A lagrange multiplier test for cross-sectional dependence in a fixed effects panel data model. Journal of Econometrics, 170(1), 164-177. Available at: https://doi.org/10.1016/j.jeconom.2012.04.004.

Bonga-Bonga, L., & Kinfack, E. (2019). The growth effect of trade openness on African countries: Evidence from using an instrumental variable panel smooth transition model. MPRA Paper No. 92111. Retrieved from: https://mpra.ub.uni-muenchen.de/92111/.

Breusch, T. S., & Pagan, A. R. (1980). The Lagrange multiplier test and its applications to model specification in econometrics. The Review of Economic Studies, 47(1), 239-253. Available at: https://doi.org/10.2307/2297111.

Dar, A., & Amirkhalkhali, S. (2003). On the impact of trade openness on growth: Further evidence from OECD countries. Applied Economics, 35(16), 1761-1766. Available at: https://doi.org/10.1080/0003684032000129020.

Dritsaki, C. (2015). Effect of trade openness and foreign direct investment on economic growth: An empirical research of the three Baltic countries. International Research Journal of Finance and Economics, 133, 191-199.

Dumitrescu, E.-I., & Hurlin, C. (2012). Testing for Granger non-causality in heterogeneous panels. Economic Modelling, 29(4), 1450-1460. Available at: https://doi.org/10.1016/j.econmod.2012.02.014.

Granger, C. (1969). Investigating causal relations by econometric models and cross-spectral methods. Econometrica, 37(3), 424-438. Available at: https://doi.org/10.2307/1912791.

Gries, T., & Redlin, M. (2012). Trade openness and economic growth: A panel causality analysis. Center For International Economics, Working Paper Series, No. 2011-06, June 2012.

Grossman, G. M., & Helpman, E. (1991). Innovation and growth in the global economy. Cambridge, MA: MIT Press.

Guei, K. M., & Roux, L. P. (2019). Trade openness and economic growth: Evidence from the economic community of Western African States region. Journal of Economic and Financial Sciences, 12(1), 1-9. Available at: https://doi.org/10.4102/jef.v12i1.402.

Im, K. S., Pesaran, M. H., & Shin, Y. (2003). Testing for unit roots in heterogeneous panels. Journal of Econometrics, 115(1), 53-74. Available at: https://doi.org/10.1016/s0304-4076(03)00092-7.

Levin, A., Lin, C.-F., & Chu, C.-S. J. (2002). Unit root tests in panel data: Asymptotic and finite-sample properties. Journal of Econometrics, 108(1), 1-24. Available at: https://doi.org/10.1016/s0304-4076(01)00098-7.

Levine, R., & Renelt, D. (1992). A sensitivity analysis of cross-country regressions. American Economic Review, 82(4), 942-963.

López, R. A. (2005). Trade and growth: Reconciling the macroeconomic and microeconomic evidence. Journal of Economic Surveys, 19(4), 623-648. Available at: https://doi.org/10.1111/j.0950-0804.2005.00264.x.

Lucas, R. E. (1988). On the mechanics of economic development. Journal of Monetary Economics, 22(1), 3-42.

Mangir, F., Kabaklarlı, E., & Ayhan, F. (2017). An analysis for the relationship between trade openness and economic growth: Evidence for ten African countries. Journal of Management and Economics Research, 15, 58-71. Available at: https://doi.org/10.11611/yead.373442.

Muhammad, F., & Jian, Z. (2016). The relationship between trade openness and economic growth in Muslim countries: An empirical investigation. Economics, 5(2), 15-19. Available at: https://doi.org/10.11648/j.eco.20160502.11.

Pesaran, M. H. (2007). A simple panel unit root test in the presence of cross-section dependence. Journal of Applied Econometrics, 22(2), 265-312. Available at: https://doi.org/10.1002/jae.951.

Pesaran., M. H. (2004). General diagnostic test for cross section dependence in panels. Cambridge Working Papers in Economics 0435, Faculty of Economics, University of Cambridge. Retrieved from: http://www.econ.cam.ac.uk/research-files/repec/cam/pdf/cwpe0435.pdf .

Ramsey, F. P. (1928). A mathematical theory of saving. The Economic Journal, 38(152), 543-559. Available at: https://doi.org/10.2307/2224098.

Romer, P. M. (1986). Increasing returns and long-run growth. Journal of Political Economy, 94(5), 1002-1037. Available at: https://doi.org/10.1086/261420.

Silajdzic, S., & Mehic, E. (2018). Trade openness and economic growth: Empirical evidence from transition economies. In Book: Trade and Global Market, Chapter 2. 9-23. Publish intechOpen, Retrieved from: http://dx.doi.org/10.5772/intechopen.75812 .

Solow, R. M. (1956). A contribution to the theory of economic growth. The Quarterly Journal of Economics, 70(1), 65-94.

Zeren, F., & Ari, A. (2013). Trade openness and economic growth: A panel causality test. International Journal of Business and Social Science, 4(9), 317-324.

Views and opinions expressed in this article are the views and opinions of the author(s), Asian Economic and Financial Review shall not be responsible or answerable for any loss, damage or liability etc. caused in relation to/arising out of the use of the content.