Economic Growth, Energy Consumption, and Trade Openness Nexus: Evidence from Net Energy Importing Middle-Income Countries

Abstract

The study explores both the long- and short-run liaisons between three conceptual dimensions: economic growth, energy consumption, and trade openness in 29 net energy importing middle-income economies using annual data from 1990 to 2019. We hereby assess ARDL models which examine the long-run links by integrations in between these three conceptual variables, and additionally Dumitrescu-Hurlin and Granger causality tests for panel and individual country models, respectively. For panel country samples, we reveal bidirectional causality connection between trade openness and economic growth along with unidirectional causalities from economic growth to energy consumption and from energy consumption to trade openness in the short-run. Bidirectional positive feedback relationships stand between economic growth - energy consumption and trade openness - energy consumption in the full sample and the upper middle-middle economies subsample in the long-run. Findings for individual country estimations reveal significant long-run relationships between energy consumption and economic growth in 12, energy consumption and trade openness in 6, and economic growth and trade openness in 9 of the middle-income economies examined.

Headnote

Summary: The study explores both the long- and short-run liaisons between three conceptual dimensions: economic growth, energy consumption, and trade openness in 29 net energy importing middle-income economies using annual data from 1990 to 2019. We hereby assess ARDL models which examine the long-run links by integrations in between these three conceptual variables, and additionally Dumitrescu-Hurlin and Granger causality tests for panel and individual country models, respectively. For panel country samples, we reveal bidirectional causality connection between trade openness and economic growth along with unidirectional causalities from economic growth to energy consumption and from energy consumption to trade openness in the short-run. Bidirectional positive feedback relationships stand between economic growth - energy consumption and trade openness - energy consumption in the full sample and the upper middle-middle economies subsample in the long-run. Findings for individual country estimations reveal significant long-run relationships between energy consumption and economic growth in 12, energy consumption and trade openness in 6, and economic growth and trade openness in 9 of the middle-income economies examined.

Keywords: Economic growth, Energy consumption, Trade openness, Middleincome countries, Cointegration, Causality.

JEL: C22, C23, F14, Q43, 050.

(ProQuest: ... denotes formulae omitted.)

Energy is a critical resource and a vital input factor in economic production activities as it eventually stimulates economic growth. The widely accepted argument for net energy importing low-income and middle-income countries, which are relatively more sensitive to fluctuations in energy prices, is that large amounts of energy import payments create unrelenting macroeconomic stress on these financially unstable economies. Therefore, understanding the economic growth-energy consumption nexus is especially essential for net energy importing low-income and middle-income countries, where the need to import energy may naturally pose unique economic challenges. In this context, the first intention of this study is to contribute to the economics literature by considering the link between energy consumption and economic growth focusing on net energy importing middle-income countries. The other phenomenon, which is frequently examined in the economics literature, is the economic growth and trade openness nexus. The effect of trade openness on growth can be either positive or negative especially in net energy importing low-income and middle-income countries. The second intention of this study is to explore the liaison between trade openness and economic growth in net energy importing middle-income countries. Whether the increase in trade causes an increase or a decrease in energy demand in net energy importing low-income and middle-income countries is another empirical question. However, the association between these variables in net energy importing low-income and middle-income countries remain as an understudied area. The last intention of this study is then to scrutinize the trade openness-energy consumption nexus in net energy importing middle-income countries.

We aim to add value to the existing literature by scrutinizing the liaisons in between economic growth, energy consumption, and trade openness with a particular focus on the net energy importing middle-income countries and to the potential policy implications which will add up the sustainability of their economic growth in the first place.

1. Conceptual Framework

It is prevalent in related literature that economic growth (ecgrw) drives energy consumption (encon). therefore energy consumption may determine the levels of economic growth as well. There are multiple assertions on the link between economic growth and energy consumption in the recent economics literature. Empirical studies performed to determine the causal relationship between ecgrw and encon are based on four hypotheses, namely growth, conservation, neutrality, and feedback hypothesis (James E. Payne 2010; Ilhan Ozturk and Ali Acaravci 2013). The unidirectional causality running from encon to ecgrw is discussed under the growth hypothesis where encon plays an important role in stimulating economic growth. Policies aiming to reduce energy consumption may have an adverse effect on ecgrw. The other unidirectional causality, this time from ecgrw to encon. is implied as the conservation hypothesis. In this hypothesis, it is argued that energy conservation policies will have no effect on ecgrw. The neutrality hypothesis assumes the absence of causal relationship between ecgrw and encon. Finally, the feedback hypothesis states the presence of a bidirectional causality between two variables. This hypothesis states that the reduction in energy supply for any reason will have a negative impact on the ecgrw which will result in lower energy demand (Emrah Beşe and Salih Kalayci 2021).

Trade openness (trdop) has a positive contribution to economic growth in several ways. Through technological progress and factor endowment, trdop enables a country to specialize in producing goods for which it has comparative advantage. Another expected effect of trdop is the increase in the market size and the lead in competition that allows for the efficient allocation of scarce resources. Moreover, trdop enables the transfer of managerial expertise and technical know-how from one country to another. In addition, investors from abroad can play an important role in providing positive net factor payments by participating in projects in the domestic country (Alexander Bilson Darku and Richard Yeboah 2018). Numerous theoretical and empirical studies in economics literature reveal that the expansion of trade improves economic growth indicators of an economy contributing to poverty reduction. Trade openness, which also increases energy use, is therefore considered a stimulus for economic activity (Perry Sadorsky 2012). However, trdop might have negative effects on economic growth in the economies which are not able to manufacture high quality products, or which have economies based on agriculture (Ricardo Haussmann, Jason Hwang, and Dani Rodrik 2007; Yaya Keho 2017).

The increase in international trade volume that emerges alongside openness in a country also upsurges economic activities and energy demand. Energy is eventually an essential input for machinery and equipment in the production process. Moreover, transportation of raw materials, parts, and finished products from one country to another requires energy. Inadequacy of energy supply may adversely affect trdop (Sadorsky 2012; Muhammad Shahbaz, Saleheen Khan, and Mohammad Iqbal Tahir 2013). The economic conditions of a country, the size of the ecgrw, and the trdop nexus are also among the determinants of the effect of trdop on encon. Theoretically, trdop impacts encon, whereas the feedback effect is not obvious. Trade openness and energy demand nexus emerges in terms of scale effect, technical effect, and composite effect. The scale effect is commonly referred to as the increase in domestic production leading to a shift in energy demand due to the expanded production. The determinants of the impact of trdop on encon are the economic conditions of the country and the association degree of the link between trdop and ecgrw. Trade openness enables developing economies to import modern technologies that will positively impact the manufacturing industry by lowering energy intensity and enabling more output to be produced. The transformation based on advanced technology in manufacturing is generally considered as the technical effect. The shift from agriculture to industry and from industry to services in an economy denote the composite effect. The increase in energy consumption is relatively less during the developments in the agricultural sector, which is accepted as the initial stages of economic development. As economic development shifts from agriculture to industry, the energy consumption level increases. Finally, economic development evolves from industry to services when encon is less (Shahbaz et al. 2014).

2. Empirical Literature

Although there has been extensive literature produced on the causal association between ecgrw and encon conflicting research results have emerged on the direction of causality. Jude C. Eggoh, Chrysost Bangake, and Christophe Rault (2011) focused on 21 African countries and their subgroups as net energy importing and net energy exporting economies. Panel test results of their study reveal a cointegrating relationship between real GDP and encon for both subgroups as well as for the whole sample. Kosta Josifidis, Radmila Dragutinović Mitrovič, and Olgica Ivančev (2012) explored the heterogeneity of growth and estimated the growth determinants in the Western Balkan and Emerging European economies. The results indicated that trade openness had played an important role in boosting economic growth in both groups of countries from 1997 to 2009. Abdul Jalil (2014) investigated the association of encon and ecgrw in 19 net energy exporting and 29 net energy importing countries from 1970 to 2012 and observed unidirectional causality running from encon to ecgrw for net energy importing countries sample. Ömer Esen and Metin Bayrak (2017) examined the impacts of encon on ecgrw by using the data of 75 net energy-importing countries from 1990 to 2012. Evidence from the panel data and some individual country estimations reveal significantly positive log-run associations between encon and ecgrw. Afees Salisu et al. (2018) studied the relationship between encon and ecgrw in five oil exporting and six oil importing countries using annual data spanning throughout the years 1980-2014. The findings of the Panel ARDE estimations indicate a significantly positive influence of ecgrw on encon in the long-run both for the oil exporting and oil importing countries samples. Obindah Gershon, Nnaemeka Emmanuel Ezenwa, and Romanus Osabohien (2019) examined the impacts of oil price shocks on net oil-importing developing countries from 1980 to 2015. The findings obtained from the Granger causality test indicate that oil prices causes GDP per capita in Liberia and Sierra Leone. From the empirical analysis based VAR model and Impulse response, they found that, the increase in oil prices temporarily increases GDP per capita in the examined countries. Adem Üzümcü, Ülker Çam-Karakaş, and Adem Karakaş (2019) explored the association between the change in energy import and ecgrw within the context of energy consumption in the sample of 23 net energy importer developed and developing countries by employing panel cointegration and Granger causality tests. They found that a change in energy import increases ecgrw in a way that supports growth hypothesis, and the existence of bilateral causality between two variables confirming feedback hypothesis. Jeffrey Kouton (2019) analyzed the asymmetric heterogeneous association of encon and ecgrw by using the data of 19 African countries from 1971 to 2014. Empirical results indicate the existence of asymmetric association in the series both in the long-run and the short-run.

Although there are various studies investigating the relationship between ecgrw, encon, and trdop in net energy importing or developing countries, there is still a lack in the literature particularly focusing only on net energy importing low-income and/or middle-income economies. However, some country-specific or country-group studies presented in Table 1 explicitly include some of these countries.

The studies examining the tripartite association between ecgrw, encon, and trdop, which are few in the literature, are mainly based on three assumptions: the existence of bidirectional causality between any of the two variables; the presence of a unidirectional causality running from one of the variables to a variable remained; and non-existence of causal relationship between the variables (Sofien Tiba and Mohamed Frikh 2018). Ozturk and Acaravci (2013) explored the causal association between economic growth, energy consumption, trade, financial development, and carbon emissions in Turkey. The findings of the bounds F-test indicate the existence of long-run association between energy consumption,per capita real income, and openness. Phouphet Kyophilavong et al. (2015) analyzed the relationship between ecgrw, encon, and trdop in Thailand using annual data from 1971 to 2012. Bayer and Hanek cointegration test results of their study reveal that encon and trdop stimulate ecgrw in the long-run. The causality tests indicate that bidirectional causality exists between ecgrw and encon, and between encon and trdop in Thailand. Ramphul Ohlan (2018) tested the association between ecgrw and encon in India spanning from 1971 to 2016. The empirical estimations provide evidence that ecgrw is stimulated by encon in India.

3. Research Methodology

3.1 Data

This empirical study examines annual time serial data on net energy importing middle-income economies over the period 1990-2019. According to the Umar Serajuddin and Nada Hamadeh 2020), 29 economies are lower income, 50 are lower middle-income, and 56 are upper middle-income. Low-income countries are not included in the study due to the absence of available data. The World Bank data is used to determine the net energy importing countries. Based on data availability, the final dataset includes 29 middle-income countries of which 11 are lower middle-income (Bangladesh, Benin, Côte d'Ivoire, El Salvador, Ghana, Honduras, India, Kyrgyz Republic, Pakistan, Ukraine, Zimbabwe) and 18 are upper middle-income economies (Albania, Belarus, Brazil, Bulgaria, China, Costa Rica, Cuba, Dominican Republic, Guatemala, Jamaica, Jordan, Lebanon, Namibia, Panama, Paraguay, Peru, Thailand, Turkey). Data relating to economic growth, energy, trade, capital, and labor was obtained from the World Development Indicators database of the World Bank (2021)1 (Table 2).

3.2 Estimating Models

Theoretical discussions and the findings of econometric studies summarized in the previous section reveal evidence that energy consumption and trade openness have roles in stimulating economic growth. We assume that the Cobb-Douglas production function (C-D) can capture the relationship between these variables. The most standard form of the C-D is:

... (1)

where Y, A, K, L and e represent the total production, total factor productivity, capital, labor, and white noise, respectively. The model is expanded to include encon and trdop by taking inference from theories and existing empirical studies asserting that energy consumption and trade play vital roles in economic growth, both directly and as a complement to labor and capital in the production process.

... (2)

The final augmented C-D of the study is constructed as in Equation (3) by including ecgrw, gfcaf and Ibfop instead of У, AK, and L, respectively, where variables are converted into natural logarithms:

... (3)

where ßQ= /пЛ0 is the constant term, ßt (z = 1, 2, 3, 4, 5) is the parameter that needs to be estimated, u and t represent the error term and the year respectively, г indicates the country in single country estimations and the country index in panel data form. Within the framework of the ARDE approach, the Equation (3) is modelled for each dependent variable as:

Model-1

... (4)

Model-2

... (5)

Model-3

... (6)

The panel ARDL representations of the models are formulated as follows:

Model-1

... (7)

Model-2

... (8)

Model-3

... (9)

3.3 Econometric Methodology

Testing for unit root in the data is considered as the natural start of any cointegration or causality analyses method using time series. Augmented Dickey Fuller (ADF) and Phillips-Perron (PP) conventional unit root tests were performed to examine the stationarity of the individual country data series. However, applying the stationary method without considering the cross-sectional dependence may cause biased estimation analysis in panel data. Therefore, it is important to conduct the cross-section dependence (C-D) test for panel datasets (M. Hashem Pesaran 2006). A second-generation panel unit root test (CIPS), and Levin-Lin-Chu, (LLC), Im-Pesaran-Shin (IPS), Fisher-ADF and Fisher-PP first-generation panel unit root tests were utilized to determine the integration order of variables in panel settings. Another requirement is the evaluation of optimal lag lengths since autoregressive models are sensitive to the number of lags. This study chose the optimal length lags by means of the Schwarz information criterion.

ARDL bounds test F-statistic was used to see whether a cointegration relationship existed between variables in the individual country models. The ARDL cointegration approach of Pesaran and Yongcheol Shin (1999) and Pesaran, Shin, and Richard J. Smith (2001), which has some advantages over conventional cointegration techniques, was preferred to capture the long-run association between variables. One of the major benefits of this approach is that the variables may have different optimal lags and it employs a single reduced form equation. Furthermore, entirety of the series in the model does not need to be of equal order of integration. However, the critical values for bound test to cointegration (Pesaran, Shin, and Smith 2001; Paresh Kumar Narayan 2005) will not be valid if the order of integration of one of the variables to be used in the analyses is two or greater. When calculating the first bound, it was assumed that all variables in the ARDL model were integrated of zero order (1(0)). The calculation of the second bound is done under the assumption that the variables are integrated in the first order (1(1)). The robustness of the estimated individual country ARDL models were examined by utilizing Breusch-Godfrey to test for autocorrelation in the errors in the models by Breusch-Pagan-Godfrey to measure how errors increase across the explanatory variables, by Jarque-Bera to confirm normality, and by CUSUM and CUSUM square to monitor the stability of parameters. For panel ARDL models, the Hausman Test (Jerry A. Hausman 1978) was conducted to decide between the Mean Group (MG) and Pooled Mean Group (PMG) estimators. PMG estimations are considered as more efficient than MG if the parameters are homogeneous. The h0 of the Hausman Test between MG and PMG is that both MG and PMG are consistent, while MG is inefficient against the Нг of PMG being consistent. In other words, MG is used in panel ARDL estimations if the outcome of the Hausman Test gives a /?-value greater than 0.05, while PMG is preferred if the test statistic is significant at 0.05 level.

Finally, causality tests were employed to capture the direction of the association between series in the short-run. If the findings of cointegration tests reveal long-run equilibrium relationships, then there must be a unidirectional or a bidirectional causality between the variables (Robert F. Engle and Clive W. J. Granger 1987). Granger Pairwise Causality test was employed to see the cause and effect between ecgrw, encon, and trdop series of individual countries. For panel series, Dumitrescu and Hurlin causality test was preferred. The methodology proposed by Elena-Ivona Dumitrescu and Christophe Hurlin (2012) proceeds by testing the Homogeneous Non-Causality hypothesis where h0 is rejected if a causal link exists between two variables in the cross-section units of the panel.

4. Empirical Results

4.1 Panel Models

Before employing unit root tests, the presence of cross-sectional dependence in panels was explored by utilizing the C-D test. Based on the results (Appendix Table Al), the null hypothesis of no cross-sectional dependence was rejected in all models. Therefore, CIPS panel unit root test was conducted alongside the first-generation unit root tests LLC, IPS, ADF-Fisher and PP-Fisher. The findings revealed that the variables were 1(0) or 1(1) and did not have order two of integration (Appendix Table A2). Furthermore, Hausman test was employed to see whether group means were consistent. The test results for Panel-1 (/2= 28.196,p = 0.042), Panel-2 (y2 = 22.016,p = 0.044), and Panel-3 (x2 = 31.046, p = 0.000) revealed that PMG was consistent for panel ARDL estimations. Alongside the Hausmann Test, we performed PMG-ARDL in which h0 of no cointegration against of cointegration was tested to capture the existence of the cointegrated combination of the series. The findings of PMG-ARDL estimations (Table 3) show that ecgnv is significantly and positively influenced by both encon, and trdop in Panel-1 and Panel-3 in the long-run. However, for Panel-2, which includes lower middle-income countries, it was observed that neither encon nor trdop have significant impact on ecgnv in the long-run. In Model-2, where encon is the dependent variable, the long-run effect of ecgnv on encon is significant and positive in all panels as it was expected. The impact of trdop on encon was significantly negative in Panel3. It was determined from the Panel-1 PMG-ARDL estimations that trdop is significantly negatively affected by ecgnv in the long-run. For all the panels, test results show that encon had significantly positive effect on trdop in the long-run.

Finally, the Dumitrescu and Hurlin heterogeneous panel causality tests were employed on each panel to capture the short-run causal relationship between ecgnv, encon, and trdop. The findings indicate unidirectional causality running from ecgnv to encon and running from encon to trdop for all-panel samples. It was revealed that bidirectional causality exists between ecgnv and trdop for net energy importing upper middle-income countries sample and for the full sample, while no significant causality relationship was captured between ecgnv and trdop for the sample including lower middle-income economies. However, the results do not validate causality running from encon to ecgnv and from trdop to encon for any of the panel samples in the short-run (Table 4).

4.2 Individual Country Models

The ARDL cointegration approach has the advantage of incorporating 1(0) and 1(1) variables in the same estimation. However, if any variable has second-order integration, the generated F-statistics will be invalid in examining the cointegration relationship. In this context, we first explored the stationarity of the series of each country via the ADF and PP tests. The findings imply that the variables do not have second-order integration except the series of Bangladesh, Belarus, China, Cote d'Ivoire, Cuba, Ghana, Kyrgyz Republic, and Ukraine (Appendix Table A3). Following unit root tests, the estimated equations were tested to see how robust they were. Results reveal that individual country models passed diagnostic tests except Benin, Bulgaria, Costa Rica, Dominican Republic, El Salvador, Jamaica, Lebanon, Namibia, Pakistan, and Peru in Model-1; Benin, Lebanon, Namibia, and Panama in Model-2; and Benin, Costa Rica, Dominican Republic, Honduras, India, Lebanon, Panama, Peru, and Zimbabwe in Model-3. Diagnostic test findings are presented in Appendix Table A4. The charts showing the results of CUSUM and CUSUM-Square tests within the critical bounds of 5 percent level of significance are available on request from the corresponding author.

An ARDL Bounds test was employed on the models to see the cointegration association between the series for the models: (i) which had variables to have different integration order 1(0), 1(1), or a combination of both; and (ii) passed diagnostic tests. The F-statistic of Model-2 for Bulgaria and Costa Rica, and Model-3 for Brazil and Paraguay, which were falling below the lower bound, indicate that cointegration is not possible for these models of the above-mentioned countries. The F-statistic of Model1 1 Guatemala and Thailand; and Model-2 for El Salvador, Jamaica, and Jordan; Model-3 for Jordan was falling between the bounds, so conclusion was made that cointegration in these models is inconclusive. The F-statistic of the rest of the models was exceeding the upper bound which means that a cointegration relationship exists between series (Table 5). In other words, when ecgrw was considered as the dependent variable, a cointegration relationship was found between encon, trdop, gfcaf and Ibfop for Albania, Brazil, Honduras, India, Jordan, Panama, Paraguay, Turkey and Zimbabwe. In the model where encon was the dependent variable, a cointegration relationship between the variables was observed for Albania, Brazil, Dominican Republic, Guatemala, Honduras, India, Pakistan, Paraguay, Peru, Thailand, Turkey, and Zimbabwe. It was determined that cointegration exists between variables for Albania, Bulgaria, El Salvador, Guatemala, Jamaica, Namibia, Pakistan, Thailand, and Turkey in the model where trdop was the dependent variable.

After identifying the existence of cointegration relationships, we explored the long-run equilibrium association between the variables. The results of the ARDL bounds test presented in Table 6 show that the variables are significantly cointegrated with all cointegrating vectors: (i) between ecgrw and encon in Albania, Brazil, Honduras, India, Jordan, Pakistan, Panama, Paraguay, Peru, Thailand, Turkey and Zimbabwe; (ii) between ecgrw and trdop in Albania, Bulgaria, Guatemala, Jamaica, Pakistan, Panama, Paraguay, Thailand and Zimbabwe; (iii) between encon and trdop in Albania, Bulgaria, India, Pakistan, Thailand and Zimbabwe.

Furthermore, Granger causality tests were employed to capture the causality relationship between the variables. We found unidirectional causality running from encon to ecgrw for Panama and Thailand; from ecgrw to encon for Brazil, Honduras, and Panama; from ecgrw to encon for Brazil, Honduras and Panama; from ecgrw to trdop for Albania, Bulgaria, Jamaica and Turkey; from trdop to encon for Brazil and encon to trdop for Albania. The absence of causality running from trade openness to economic growth observed for the entirety of countries (Table 7). The findings partially confirm the results of Nicholas Apergis and Chor Foon Tang (2013), Seref Bozoklu and Veli Yilanci (2013), Shahbaz et al. (2014), Carlos Vladimir Rodriguez-Caballeroa and Daniel Ventosa-Santaulària (2016) and Haonan Zhang et al. (2021).

5. Discussion

The PMG-ARDL and Dumitrescu-Hurlin causality test results indicate the presence of cointegration and causal relationship among economic growth, energy consumption, and trade openness in all panels. Although the bidirectional positive feedback relationships stand between economic growth and energy consumption in the long-run in the full sample and in the upper middle-middle economies subsample, there are short-run unidirectional causalities running from economic growth to energy consumption in these samples. The findings for the full sample and the upper middle-middle economies subsample express that policies to reduce energy consumption in the long-run may have negative effects on economic activity and may lead to further reduction in energy consumption. The results imply divergences in the lower middle-middle economies subsample indicating a unidirectional causality running from economic growth to energy consumption both in the long- and short-run. Conservation hypothesis suggests that energy conservation policies, which may have no or little adverse effects on the economic growth is valid in the long- and short-run for net energy importing lower middle-income countries, and in the short-run for net energy importing upper middleincome economies.

Although the results show that economic growth and trade openness are impacting each other both in the short- and long-run in the full sample, the causal relationship between these two variables has been proven to be nonexistent for net energy importing lower middle-income countries. Although trade openness appears to be a significant contributor to the economic growth of upper middle-income net energy importing countries in the long-run, bidirectional causality relationship exists in the shortrun.

The results show that there is a unidirectional relationship from energy consumption to trade openness in the short-run in all panel samples. Whereas a long-run bidirectional feedback relationship stands between energy consumption and trade openness for the net energy importing upper middle-income countries sample, there is a long-run unidirectional causality showing that energy consumption stimulates trade openness in the net energy importing lower middle-income economies. Short-run unidirectional causal relationship running from economic growth to energy consumption and from energy consumption to trade openness in upper middle-income countries imply that economic growth increases the energy consumption and energy consumption boosts trade openness.

Panel data techniques are advantageous because of the increase in data points, and thus the power of statistical estimation. On the other hand, countries are not treated as independent units and differences between these countries are neglected in panel models. Because the relationship between the variables are strongly dependent on the economic structure, geographic and geological endowments, and institutional design of each country as well as the level of energy conservation and trade openness may differ between net energy importing middle-income countries, the policy implications to be proposed should be based on individual country estimations. Thus, we individually examined the relationship between economic growth, energy consumption and trade openness for net energy importing middle-income countries. Although the full panel includes 29 countries, the ARDL bounds tests for cointegration and the Granger causality tests are applied to the series of 18 net energy importing middle-income economies because the estimated models of 11 countries do not meet the requirements of robustness criteria.

The ARDL estimation findings for individual countries reveal a bidirectional relationship between the energy consumption and economic growth in the long-run in Albania, Brazil, Honduras, Turkey, and Zimbabwe. Because of the feedback effect between the economic growth and energy consumption in the long-run, we recommend that policymakers prioritize the provision of adequate and efficient energy supply by exploring and investing, inter alia, new and alternative sources of energy in these countries.

Unidirectional relationship from economic growth to energy consumption findings reveal that higher energy consumption comes with higher economic growth in India, Pakistan and Peru in the long-run, and in Brazil and Honduras in the short-run. We suggest that policymakers design energy-saving policies and take actions to support the efficient use of energy in these countries because the policies that aim to reduce the energy consumption might have little or no impact on economic growth.

The findings for Paraguay both in the short and long-run, and for Jordan in the long-run support the growth hypothesis, which means that a unidirectional causality relationship exists, running from energy consumption to economic growth. This hypothesis considers energy use as one of the engines of economic growth. The explicit unidirectional causality implies that policies aimed at reducing energy consumption may have negative effects on economic growth in Paraguay and Jordan.

The findings of this study reveal a reverse association between the economic growth and energy consumption in the short- and long-run in Panama and Thailand. The unidirectional causality from energy consumption to economic growth exists in Panama in the long-run, and economic growth Granger causes energy consumption in the short-run, but vice versa is observed in Thailand. The results imply the necessity of comprehensive analysis of the potential consequences of energy-oriented policies to be implemented in Panama and Thailand. We suggest that these countries create and apply flexible and sector-specific energy policies.

The neutrality hypothesis which considers that there is no causal relationship between the energy consumption and economic growth is valid for Dominican Republic and Guatemala. Therefore, energy conservation policies might not affect economic growth, and economic growth does not necessarily cause an improvement of the energy sector in these countries.

The findings show that trade openness does not Granger cause energy consumption in any of the net energy importing middle-income countries. However, The ARDL bounds test findings bear significant unidirectional relationship running from trade openness to energy consumption in Albania, India, Thailand, and Zimbabwe. The results imply that, promoting international trade does not significantly increase energy consumption, and designing energy conservative policies may not reduce international trade in the short-run among the net energy importing middle-income countries explored. Nevertheless, promoting international trade significantly increases energy consumption in India and Thailand in the long-run, while reducing it in Albania and Zimbabwe. Policy makers in these economies should consider the long-run significant impact of trade openness on energy consumption when formulating and implementing trade and energy policies. On the other hand, a unidirectional causality running from energy consumption to trade openness is found in case of Bulgaria, where energy consumption boosts trade significantly and positively in the long-run. The findings depict that designing energy conservative policies may reduce international trade in Bulgaria in the long-run.

The results validate the presence of a feedback effect in Pakistan, as energy consumption and trade openness are interdependent in the long-run. The findings indicate that promoting international trade enhances energy consumption, and implementing energy conservative policies may lead to a reduction in international trade. The feedback relationship result suggests that Pakistan should take international trade into account for the future energy prediction efforts and adopt energy expansion policies.

Although the results appear to be robust, some of the net energy importing middle-income countries do not have a complete series for the variables to be included in the panel settings. Because of the absence of available data for these countries, the precision of the panel estimations remains imperfect. The trade openness data, which consists of volume based aggregate foreign trade data may be misleading for assessing the net impact on energy consumption and/or economic growth. Most of the middlelevel income countries have import-dependent production activities that end up with exports because of trans-trade activities and international logistics. Moreover, the choice in the mode of production is mainly determined by the local availability of raw materials and energy, as well as by the environmental regulations in the country and policies regarding their implementation (Salih Kalayci and Cihan Özden 2021). Also, one may argue that both the demand and supply sides affect inflation, which are normally found in the lower- and middle-income group of countries may negatively affect economic growth and trade activities eventually through the increasing cost of energy.

The political and macroeconomic circumstances of economies should also be considered for such assessment because of their risky nature. Although the relatively higher interest rates and devaluation potential of the domestic currency against foreign currencies cover such risks along with the relatively higher debt position, policy-makers must concentrate on how to increase trade openness and how to reach a sustainable economic growth in setting the objectives, which are also the prevailing solutions to shift the level of the economy. In addition, electricity theft and loss rates are high as well the excise tax rates on fuel in some low- and middle-income countries. This is one of the reasons for the cost paradox of energy confining their economic growth in the ongoing high inflation. Furthermore, lower investments are the results of a low propensity to save in these low- and middle-income economies, where a lower concentration of foreign direct investment is most likely to occur because of incompetent sub-structures, a weak legal system, and the lack of trustful politics in meeting future expectations and stability. However, any policies to cope with such problems can still attain an economic growth level that is high enough with respect to their high population growth and low employment rates. Therefore, the results of this study will expectedly help policy authorities to integrate convenient strategies to develop and promote country-level solutions on the main variables of the study.

6. Conclusion

This study has explored the long-run equilibrium and the short-run causal association between economic growth, energy consumption and trade openness in net energy importing middle-income countries. To investigate the difference between country blocks, groups of countries were classified as lower middle-income and upper-income depending on their income characteristics in line with the World Bank classification. Based on data availability, the final dataset included 29 middle-income countries (Panel-1) 11 of which are lower middle-income (Panel-2) and 18 are upper middleincome (Panel-3) economies. We have then assessed the long-run equilibrium relationship between the economic growth, energy consumption, and trade openness of panel samples by employing the PMG-ARDL approach. The findings indicated that both the energy consumption and trade openness have positive and significant effects on the economic growth in the long-run for net energy importing middle-income economies as a whole and for net energy importing upper middle-income economies as a subgroup as well. Moreover, the findings confirm a significantly positive impact of economic growth on energy consumption for all samples examined. The PMG-ARDL estimations reveal that trade openness is significantly and negatively affected by economic growth in the long-run in net energy importing middle-income economies full sample. For all the panel samples, test results depict that energy consumption has a significantly positive effect on trade openness in the long-run. The ARDL estimation findings for individual countries reveal significant long-run relationships between energy consumption and economic growth, energy consumption and trade openness, and economic growth and trade openness in 12, 8, and 6 of middle-income economies explored, respectively.

By employing the Dumitrescu-Hurlin causality tests, we have observed a unidirectional causal relationship running from economic growth to energy consumption and running from energy consumption to trade openness in all panel samples. The findings reveal a bidirectional causality between economic growth and trade openness for the upper middle-income countries sample and the whole sample in the short-run. Economic growth is a Granger cause of energy consumption and trade openness in three and four of the middle-income economies, respectively. Granger causality test results show that there is a unidirectional causality running from energy consumption to economic growth in two, from trade openness to energy consumption in one, and from energy consumption to trade openness in two, one, and one of the net energy importing middle-income countries explored, respectively.

The findings of the study will pave the way for the lower- and middle-income countries of net energy importing nature by clearing the possibilities if any decision on a conceptual dimension will work out for any goals in attaining a sustainable economic growth, in energy strategics, and in policies coping with malignant foreign trade deficit in these economies.

Future studies to be conducted through comparative research between net energy importing low-income, middle-income, and high-income countries will be valuable. Further questions on the connections between economic growth and energy consumption, economic growth and trade openness, and energy consumption and trade openness will require dynamic general equilibrium models. Nonetheless, renewable energy sources, with an increasing part of supply in energy consumption have recently been a new direction to explore for future research in terms of value-added contributions to economic growth and trade openness.

(ProQuest: Appendix omitted.)

Footnote

1 World Bank. 2021. World Development Indicators, https://databank.worldbank.org/source/world-development-indicators (accessed September 01, 2021).

References

References

Amadou, Akilou. 2013. "Is there a Causal Relation between Trade Openness and Economic Growth in the WAEMU Countries?" International Journal of Economics and Finance, 5(6): 151-156. http://dx.doi.org/10.5539/ijef.v5n6pl51

Apergis, Nicholas, and Chor Foon Tang. 2013. "Is the Energy-Led Growth Hypothesis Valid? New Evidence from a Sample of 85 Countries." Energy Economics, 38(C): 24-31. http://dx.doi.Org/10.1016/j.eneco.2013.02.007

Appiah, Michael Owusu. 2018. "Investigating the Multivariate Granger Causality between Energy Consumption, Economic Growth and CO2 Emissions in Ghana." Energy Policy, 112: 198-208. http://dx.doi.Org/10.1016/j.cnpol.2017.10.017

Beşe, Emrah, and Salih Kalayci. 2021. "Environmental Kuznets Curve (EKC): Empirical Relationship between Economic Growth, Energy Consumption, and CO2 Emissions: Evidence from 3 Developed Countries." Pano económicas, 68(4): 483-506. http://dx.doi.org/10.2298/PAN180503004B

Bozuklu, Şeref, and Veli Yilanci. 2013. "Energy Consumption and Economic Growth for Selected OECD Countries: Further Evidence from the Granger Causality Test in the Frequency Domain." Energy Policy, 63(C): 877-881. http://dx.doi.Org/10.1016/j.enpol.2013.09.037

Chang, Tsangyao, Hsiao-Ping Chu, and Wen-Yi Chen. 2013. "Energy Consumption and Economic Growth in 12 Asian Countries: Panel Data Analysis." Applied Economics Letters, 20(3): 282-287. http://dx.doi.org/10.1080/13504851.2012.692869

Darku, Alexander Bilson, and Richard Yeboah. 2018. "Economic Openness and Income Growth in Developing Countries: A Regional Comparative Analysis." Applied Economics, 50(8): 855-869. http://dx.doi.org/10.1080/00036846.2017.1343449

Dedeoǧlu, Dinçer, and Hüseyin Kaya. 2013. "Energy Use, Exports, Imports and GDP: New Evidence from the OECD Countries." Energy Policy, 57(C): 469-476. http://dx.doi.Org/10.1016/j.enpol.2013.02.016

Dumitrescu, Elena-Ivona, and Christophe Hurlin. 2012. "Testing for Granger Non-Causality in Heterogeneous Panels." Economic Modelling, 29(4): 1450-1460. http://dx.doi.Org/10.1016/j.econmod.2012.02.014

Eggoh, Jude C., Chrysost Bangake, and Christophe Rault. 2011. "Energy Consumption and Economic Growth Revisited in African Countries." Energy Policy, 39(11): 7408-7421. http://dx.doi.Org/10.1016/j.enpol.2011.09.007

Engle, Robert F., and Clive W. J. Granger. 1987. "Co-integration and Error Correction: Representation, Estimation, and Testing." Econometrica, SSÇfY ISX-ľie. http://dx.doi.org/10.2307/1913236

Esen, Ömer, and Metin Bayrak. 2017. "Does more Energy Consumption Support Economic Growth in Net Energy-Importing Countries?" Journal of Economics, Finance and Administrative Science, 22(42): 75-98. http://dx.doi.org/10.1108/JEFAS-01-2017-0015

Gershon, Obindah, Nnaemeka Emmanuel Ezenwa, and Romanus Osabohien. 2019. "Implications of Oil Price Shocks on Net Oil-Importing African Countries." Heliyon, 5(8): 1-12. http://dx.doi.Org/10.1016/j.heliyon.2019.e02208

Gessesse, Abrham Tezera, and Ge He. 2020. "Analysis of Carbon Dioxide Emissions, Energy Consumption, and Economic Growth in China." Agricultural Economics-Czech, 66(4): 183-192. http://dx.doi.org/10.17221/258/2019-AGRICECON

Hausman, Jerry A. 1978. "Specification Tests in Econometrics." Econometrica, 46(6): 1251-1272. http://dx.doi.org/10.2307/1913827

Haussmann, Ricardo, Jason Hwang, and Dani Rodrik. 2007. "What You Export Matters." Journal of Economic Growth, 12(1): 1-25. http://dx.doi.org/10.1007/sl0887-006-9009-4

Hye, Qazi Muhammad Adnan, and Wee-Yeap Lau. 2015. "Trade Openness and Economic Growth: Empirical Evidence from India." Journal of Business Economics and Management, 16(1): 188-205. http://dx.doi.org/10.3846/16111699.2012.720587

Islam, Md. Monirul, and Md. Saiful Islam. 2021. "Energy Consumption-Economic Growth Nexus within the Purview of Exogenous and Endogenous Dynamics: Evidence from Bangladesh." OPEC Energy Review, 45(2): 191-216. http://dx.doi.org/10.! 11 l/opec.12195

Jalil, Abdul. 2014. "Energy-Growth Conundrum in Energy Exporting and Importing Countries: Evidence from Heterogeneous Panel Methods Robust to Cross-Sectional Dependence." Energy Economics, 44(C): 314-324. http://dx.doi.Org/10.1016/j.eneco.2014.04.015

Jalil, Abdul, and Mete Feridun. 2014. "Energy-Driven Economic Growth: Energy Consumption - Economic Growth Nexus Revisited for China." Emerging Markets Finance and Trade, 50(5): 159-168.

Jiranyakul, Komain. 2016. "Causal Linkages between Electricity Consumption and GDP in Thailand: Evidence from the Bounds Test." Economics Bulletin, 36(2): 921-930.

Josifidis, Kosta, Radmila Dragutinovič Mitrovič, and Olgica Ivančev. 2012. "Heterogeneity of Growth in the West Balkans and Emerging Europe: A Dynamic Panel Data Model Approach." Panoeconomicus, 59(Special Issue): 157-183. http://dx.doi.org/10.2298/PAN1202157J

Kablamaci, Baris. 2017. "A Re-examination of Causal Relation between Economic Growth and Energy Consumption: Evidence from 91 Countries." Economics Bulletin, 37(2): 790-805.

Kalayci, Salih, and Cihan Özden. 2021. "The Linkage among Sea Transport, Trade Liberalization and Industrial Development in the Context of CO2: An Empirical Investigation from China." Frontiers in Environmental Science, 9: 1-10. http://dx.doi.org/10.3389/fenvs.2021.633875

Keho, Yaya. 2017. "The Impact of Trade Openness on Economic Growth: The Case of Côte d'Ivoire." Cogent Economics and Finance, 5(1): 1-14. http://dx.doi.org/10.1080/23322039.2017.1332820

Kisswani, Khalid M. 2017. "Evaluating the GDP-Energy Consumption Nexus for the ASEAN-5 Countries Using Nonlinear ARDL Model." OPEC Energy Review, 41(4): 318-343. http://dx.doi.org/10.! 11 l/opec.12113

Kouton, Jeffrey. 2019. "The Asymmetric Linkage between Energy Use and Economic Growth in Selected African Countries: Evidence from a Nonlinear Panel Autoregressive Distributed Lag Model." Energy Economics, 83(4): 475-490. http://dx.doi.Org/10.1016/j.eneco.2019.08.006

Kyophilavong, Phouphet, Muhammad Shahbaz, Sabeen Anwar, and Sameen Masood. 2015. "The Energy-Growth Nexus in Thailand: Does Trade Openness Boost Up Energy Consumption?" Renewable and Sustainable Energy Reviews, 46: 265-274. http://dx.d0i.0rg/l 0.1016/j .rser.2015.02.004

Menyah, Kojo, Saban Nazlioglu, and Yemane Wolde-Rufael. 2014. "Financial Development, Trade Openness and Economic Growth in African Countries: New Insights from a Panel Causality Approach." Economic Modelling, 37(C): 386-394. http://dx.doi.org/10.1016/j .econmod.2013.11.044

Musila, Jacob W., and Zelealem Yiheyis. 2015. "The Impact of Trade Openness on Growth: The Case of Kenya." Journal of Policy Modeling, 37(2): 342-354. http://dx.doi.org/10.1016/jjpolmod.2014.12.001

Narayan, Paresh Kumar. 2005. "The Saving and Investment Nexus for China: Evidence from Cointegration Tests." Applied Economics, 37(17): 1979-1990. http://dx.doi.org/10.1080/00036840500278103

Ohlan, Ramphul. 2018. "The Relationship between Electricity Consumption, Trade Openness and Economic Growth in India." OPEC Energy Review, 42(4): 331-354. http://dx.doi.org/10.! 11 l/opec.12134

Ozturk, Ilhan, and Ali Acaravci. 2013. "The Long-Run and Causal Analysis of Energy, Growth, Openness and Financial Development on Carbon Emissions in Turkey." Energy Economics, 36: 262-267. http://dx.doi.Org/10.1016/j.eneco.2012.08.025

Payne, James E. 2010. "Survey of the International Evidence on the Causal Relationship between Energy Consumption and Growth." Journal of Economic Studies, 37(1): 53-95. http://dx.doi.org/10.1108/01443581011012261

Pesaran, M. Hashem, and Yongcheol Shin. 1999. "An Autoregressive Distributed Lag Modelling Approach to Cointegration Analysis." In Econometrics and Economic Theory in the 20th Century, ed. Steinar Strom, 371-413. Cambridge, M. A.: Cambridge University Press.

Pesaran, M. Hashem, Yongcheol Shin, and Richard J. Smith. 2001. "Bounds Testing Approaches to the Analysis of Level Relationship." Journal of Applied Econometrics, 16(3): 289-326. http://dx.doi.org/10.1002/jae.616

Pesaran, M. Hashem. 2006. "Estimation and Inference in Large Heterogeneous Panels with a Multifactor Error Structure." Econometrica, 967-1012. http://dx.doi.org/10.! 111/j. 1468-0262.2006.00692.x

Rodriguez-Caballeroa, Carlos Vladimir, and Daniel Ventosa-Santaulària. 2016. "Energy-Growth Long-Term Relationship under Structural Breaks: Evidence from Canada, 17 Latin American Economies and the USA." Energy Economics, 61(C): 121-134. http://dx.doi.Org/10.1016/j.eneco.2016.10.026

Sadorsky, Perry. 2012. "Energy Consumption, Output and Trade in South America." Energy Economics, 34(2): 476-488. http://dx.doi.Org/10.1016/j.eneco.2011.12.008

Salisu, Afees, Tirimisiyu Oloko, Ismail Okunoye, Ólaidé Opeloyeru, and Nafisat Olabisi. 2018. "Energy Consumption and Economic Growth in Oil Importing and Oil Exporting Countries: A Panel ARDL Approach." Centre for Econometric and Allied Research Working Paper CWPS 0048.

Serajuddin, Umar, and Nada Hamadeh. 2020. "New World Bank Country Classifications by Income Level: 2020-2021." World Bank Blogs, July 01. https://blogs.worldbank.org/opendata/new-world-bank-country-classifications-income-level-2020-2021.

Shahbaz, Muhammad, Saleheen Khan, and Mohammad Iqbal Tahir. 2013. "The Dynamic Links between Energy Consumption, Economic Growth, Financial Development and Trade in China: Fresh Evidence from Multivariate Framework Analysis." Energy Economics, 40(C): 8-21. http://dx.doi.Org/10.1016/j.eneco.2013.06.006

Shahbaz, Muhammad, Samia Nasreen, Chong Hui Ling, and Rashid Sbia. 2014. "Causality between Trade Openness and Energy Consumption: What Causes what in High-, Middle- and Low-Income Countries." Energy Policy, 70: 126-143. http://dx.doi.Org/10.1016/j.enpol.2014.03.029

Sharma, Rajesh, and Pradeep Kautish. 2019. "Dynamism between Selected Macroeconomic Determinants and Electricity Consumption in India." International Journal of Social Economics, 46(6): 805-821. http://dx.doi.org/! 0.1108/IJSE-11-2018-0586

Singh, Kamaljit, and Simmi Vashishtha. 2020. "Does any Relationship between Energy Consumption and Economic Growth Exist in India? A VAR Model Analysis." OPEC Energy Review, 44(3): 334-347. http://dx.doi.org/10.llll/opcc.12185

Tiba, Sofien, and Mohamed Frikh. 2018. "Income, Trade Openness and Energy Interactions: Evidence from Simultaneous Equation Modeling." Energy, 147(C): 799-811. http://dx.doi.Org/10.1016/j.energy.2018.01.013

Tuna, Gülfen, and Vedat Ender Tuna. 2020. "Causality Relationship between Electricity Consumption and Economic Growth: Investigation for Azerbaijan, Kazakhstan and Kyrgyzstan." International Journal of Economics, Management and Accounting, 28(1): 165-190.

Üzümcü, Adem, Ülker Çam-Karakaş, and Adem Karakaş. 2019. "Energy Import and Economic Growth: An Analysis on some Energy Importer Countries." S.C. Ü. İktisadi ve İdari Bilimler Dergisi, 20(2): 317-334.

Wang, Jianlin, Jiajia Zhao, and Hongzhou Li. 2018. "The Electricity Consumption and Economic Growth Nexus in China: A Bootstrap Seemingly Unrelated Regression Estimator Approach." Computational Economics, 52(5): 1195-1211. http ://dx. doi.org/10.1007/s 10614-017-9709-1

Xie, Hong, Yifei Cai, Chung Yan Sam, and Tsangyao Chang. 2018. "Revisit Financial Development, Trade Openness and Economic Growth Nexus in China Using a New Developed Bootstrap ARDE Test." Economic Computation and Economic Cybernetics Studies and Research, 52(4): 131-144. http://dx.doi.Org/10.24818/18423264/52.4.18.09

Yildirim, Ertugrul, Alper Aslan, and Ilhan Ozturk. 2014. "Energy Consumption and GDP in Asian Countries: Bootstrap-Corrected Panel and Time Series Causality Tests." The Singapore Economic Review, 59(2): 1-16. http://dx.doi.org/10.1142/S0217590814500106

Yildirim, Ertugrul, Deniz Sukruoglu, and Alper Aslan. 2014. "Energy Consumption and Economic Growth in the Next 11 Countries: The Bootstrapped Autoregressive Metric Causality Approach." Energy Economics, 44(C): 14-21. http://dx.doi.Org/10.1016/j.eneco.2014.03.010

Yuan, Jiahai, Yan Xu, and Xingping Zhang. 2014. "Income Growth, Energy Consumption, and Carbon Emissions: The Case of China." Emerging Markets Finance and Trade, 50(5): 169-181.

Zerbo, Eleazar. 2017. "Energy Consumption and Economic Growth in Sub-Saharan African Countries: Further Evidence." Economics Bulletin, FAff. 1720-1744.

Zhang, Haonan, Xingping Zhang, Changhong Zhao, and Jiahai Yuan. 2021. "Electricity Consumption and Economic Growth in BRI Countries: Panel Causality and Policy Implications." Emerging Markets Finance and Trade, 57(3): 859-874. http://dx.doi.org/10.1080/1540496X.2019.1601551