Causality between trade openness and energy consumption: what causes what in high, middle and low income countries.

Shahbaz, Muhammad ; Nasreen, Samia ; Ling, Chong Hui 等

ABSTRACT

This paper explores the relationship between trade openness and energy consumption using data of 91 high, middle and low income countries. The study covers the period of 1980-2010. We have applied panel cointegration and causality approaches to investigate the long run and causal relationship between the variables.

Our results confirm the presence of cointegration between the variables. The relationship between trade openness and energy consumption is inverted U-shaped in high income countries but U-shaped in middle and low income countries. The homogenous and non-homogenous causality analysis reveals the bidirectional causality between trade openness and energy consumption.

Keywords: Trade, Energy, Causality

INTRODUCTION

Trade liberalisation has affected the flow of trade (goods and services) between developed and developing countries. The Heckscher-Ohlin trade theory reveals that under free trade, developing countries would specialise in the production of those goods that are produced by relatively abundant factors of production such as labour and natural resources. Developed countries would specialise in the production of those goods that are produced by human capital and manufactured in capital-intensive activities. Trade openness entails movement of goods produced in one country for either consumption or further processing to other country. Production of those goods is not possible without the effective use of energy. Trade openness affects energy demand via scale effect, technique effect and composite effect. Other things being same, trade openness increases economic activities, thus stimulates domestic production and hence economic growth. A surge in domestic production increases energy demand , which is commonly referred as scale effect. Such scale effect is caused by trade openness. Economic condition of the country and extent of relationship between economic growth and trade openness determine the impact of trade openness on energy consumption [Shahbaz, et al. (2013); Cole (2006)]. Trade openness enables developing economies to import advanced technologies from developed economies. The adoption of advanced technology lowers energy intensity. The use of advanced technologies result in less energy consumption and more output that is usually referred to as technique effect [Arrow (1962)]. Composite effect reveals the shift of production structure from agriculture to industry with the use of energy intensive production techniques. In initial stages of economic development economy is based largely on agriculture sector, thus the use of energy is relatively less. As economy starts shifting from agriculture to industry, the energy consumption increases. Arrow (1962) calls it positive composite effect. Finally, at the later stage of economic development, economic structure shifts from industry to services, there is less energy consumption, which implies that energy intensity is lowered because of composite effect.

Energy affects trade openness via various channels. First, energy is an important input of production because machinery and equipment in the process of production require energy. Second, export or import of manufactured goods or raw material require energy to fuel transportation. Without adequate energy supply, trade openness will be adversely affected. Consequently, energy is an important input in trade expansion and adequate consumption of energy is essential to expand trade via expanding exports and imports. The relationship between trade openness and energy consumption is important. Since energy plays a key role to promote exports or imports hence policies aiming at reduction of energy consumption such as energy conservation policies will negatively impact the flow of exports or imports and hence, reduce the benefit of trade openness. The bidirectional causal relationship between trade openness and energy consumption suggests that energy expansion policies should be adopted because energy consumption stimulates trade openness and trade openness affects energy consumption [Sadorsky (2011)]. The energy conservation policies will not have an adverse effect on trade openness if causality is running from trade openness to energy consumption or if neutral relationship exists between trade openness and energy consumption [Sadorsky (2011)].

Energy consumption in the world increases parallel to technological development, increase in trade and population growth. The world average energy consumption was 1454 Kg of oil equivalent per capita in 1980, which increased to 1852 Kg of oil equivalent per capita in 2010 (see Figure 1). According to American Energy Information Administration (EIA) and the International Energy Agency (IEA), the worldwide energy consumption will on average continue to increase by 2 percent per year.

[FIGURE 1 OMITTED]

Between 1980 and 2006, energy consumption has increased but fuel consumption structure varies by region. Coal has the largest share in fuel consumption of the world, accounting for 30.4 percent of total increase; Asia and Oceania contributed 97.7 percent of total coal increase between 1980 and 2006. During the same period, natural gas ranks second in total energy consumption, accounting for 28.7 percent, Asian and Oceania still contributed the largest part, 24 percent of total gas increase, Eurasia, Europe and Middle East contributed about 17 percent and 20 percent respectively. Oil ranked as the third fuel in total consumption, accounting for 21.5 percent. Asia and Oceania still were the biggest contributors; accounting for about 67.9 percent of increase in oil consumption. The nuclear power contributed about 10.7 percent to total increase, the increase was mainly contributed by Europe, North America and, Asia and Oceania where more new nuclear reactors have been started. Hydropower has developed in Asia and Oceania and Central and, South America, because of their abundant hydro resources. And these two regions contribute 80 percent to global hydropower increase. However, global industry sector has reduced the use of total energy from 33 percent in 1980 to 27 percent in 2006 because most developed countries used less energy in industry by improvement in energy efficiency, technology development and major production structure changes.

Growth in world energy consumption reached 5.6 percent in 2010, the highest growth rate since 1973. Energy consumption in OECD countries grew by 3.5 percent while in non-OECD countries by 7.5 percent in 2010. Chinese energy consumption grew by 11.2 percent and China surpassed the United States as the world's largest energy consumer. Oil remained the world's leading fuel in 2010, and accounted for 33.6 percent of global energy consumption. World natural gas consumption grew by 7.4 percent in 2010, the most rapid increase since 1984. The United States witnessed the world's largest increase in consumption, which rose by about 5.6 percent in 2010. Asian countries also registered large increase of about 10.7 percent, led by a 21.5 percent increase in India. Coal consumption grew by 7.6 percent in 2010, the fastest global growth since 2003. The share of coal in world energy consumption is 29.6 percent, more than 25.6 percent of ten years ago. China consumed 48.2 percent of world coal and accounted for nearly two-third of global coal consumption. The use of modern renewable energy sources including wind, solar, geothermal, marine, modern biomass and hydro continued to grow rapidly and accounted for 1.8 percent of world energy consumption in 2010, up from 0.6 percent in 2000. Energy use in transport sector increased very rapidly during the recent years due to rapid economic development and population growth. Over the past 30 years, energy use in transport sector has doubled. Transport sector accounts for 25 percent of world energy consumption in 2010 [International Energy Agency (2012)].The volume of merchandise trade among countries has been rapidly increasing for last two decades due to globalisation. Global merchandise trade (exports plus imports of goods) was US$ 3.8 trillion in 1980 but it amounted to US$ 37 trillion in 2010 (see Figure 2).

[FIGURE 2 OMITTED]

In 2006, merchandise exports in volume terms increased among regions. Exports from North America and Asia grew faster than imports. The growth rate of Asian export was 13 percent while imports grew by 9 percent. Europe recorded balanced export and import growth of 7 percent. For South and Central America, the Commonwealth of Independent States, Africa and the Middle East, import growth was larger than exports. This pattern is attributed to more favourable terms of trade due to increases in commodity prices in the past few years. The global economies faced negative trade shock in 2009. This negative trade shock was mainly due to massive contraction of global demand that reduced commodity prices in all regions of the world. The trade shock was strongest in transition economies and the economies of Western Asia and Africa. However, the similar situation does not exist in 2010. All WTO regions experienced double-digit increase in the dollar value of both exports and imports in 2010 due to rise in prices of fuel and other commodities. The top merchandise exporters in 2010 were China (US$ 1.58 trillion) followed by United States (US$ 1.28 trillion), Germany (US$ 1.27 trillion), Japan (US$ 770 billion) and Netherlands (US$ 572 billion). The leading merchandise importers in 2010 were United States (US$ 1.97 trillion), China (US$ 1.40 trillion), Germany (US$ 1.07 trillion), Japan (US$ 693 billion) and France (US$ 606 billion) (Source: World Trade Report, 2011).

There are a few studies that examined the relationship between energy consumption and economic growth [Masih and Masih (1996); Yang (2000); Narayan, et al. (2008)], energy consumption and exports [Narayan and Smyth (2009); Lean and Smyth (2011); Halicioglu (2010); Shahbaz, et al. (2013a)]. However, the relationship between trade openness and energy consumption is still understudied. The objective of this study is to fill this gap by investigating the relationship between trade openness and energy consumption using global data of 91 high, middle and low-income countries for the period 1980-2010. The pooled mean group and mean group models are used to show non-linear relationship between trade openness and energy consumption. Test for establishing the long-run relationships between variables are carried out by using the panel cointegration approach developed by Larsson et al. (2001) while test for causality is conducted by using a modified version of Granger causality test developed by Hurlin and Venet (2001).

The rest of the paper is organised as follows: Section 2 gives a brief review of empirical studies, Section 3 presents the methodology and data source, Section 4 presents the results and discussion and Section 5 gives the conclusions and policy implications.

2. LITERATURE REVIEW

There is an extensive literature available on the relationship between economic growth and energy consumption. Energy consumption is an important factor of production like capital and labour and it affects economic growth. After the end of 1970s energy crisis, many studies [e.g. Kraft and Kraft (1978), Akarca and Long (1979 and 1980), Yu and Choi (1985)] exposed that energy consumption is positively correlated with economic growth. However, empirical evidence provided by Zahid (2008), Amirat and Bouri (2010), Noor and Siddiqi (2010), Apergis and Payne (2010) is conflicting about direction of causality. For instance, Nondo and Kahsai (2009) investigated the long-run relationship between total energy consumption and economic growth for a panel of 19 African countries. They applied Levine, et al. (2005), Im, et al. (2003) and Hadri (2005) panel unit root tests to test the integrating properties of real GDP and total energy consumption. Their analysis indicated that both the variables are cointegrated for long run relationship confirmed by Pedroni (1999) panel cointegration approach. Moreover, they noted that economic growth is cause of energy consumption in long run as well as in short run. Noor and Siddiqi (2010) investigated the causal relationship between per capita energy consumption and per capita GDP in five South Asian countries namely Bangladesh, India, Nepal, Pakistan and Sri Lanka. They applied panel unit root tests IPS, LLC and MW, and Pedroni cointegration as well as Kao residual cointegration approaches. They reported that energy consumption enhances economic growth. Their causality analysis reveals that economic growth Granger causes energy consumption in South Asian countries. (1)

There are a few studies investigating the relationship between trade openness and energy consumption. For instance, Cole (2006) examined the relationship between trade liberalisation and energy consumption. Cole (2006) used data of 32 countries and found that trade liberalisation promotes economic growth, which boosts energy demand. Moreover, trade liberalisation stimulates use of capital intensive techniques, which in turn affects energy consumption. Jena and Grote (2008) investigated the impact of trade openness on energy consumption. They noted that trade openness stimulates industrialisation via scale effect, technique effect, composite effect and comparative advantages effect, which affect energy consumption. Narayan and Smith (2009) examined the causal relationship between energy consumption and economic growth by incorporating exports as an indicator of trade openness in production function for a panel of six Middle Eastern countries namely Iran, Israel, Kuwait, Oman, Saudi Arabia and Syria. They applied panel unit root test, panel cointegration and panel causality tests. Their analysis confirmed the presence of cointegration relationship between variables. Furthermore, they reported that that a short-run Granger causality exists running from energy consumption to real GDP and from economic growth to exports but neutral relationship is found between exports and energy consumption.

Later on, Sadorsky (2011) examined the causal relationship between total energy consumption and trade openness. The panel means group cointegration and panel Granger causality approaches were used for the panel of 8 Middle Eastern countries namely, Bahrain, Iran, Jordan, Oman, Qatar, Saudi Arabia, Syria and UAE. The empirical evidence reported that long run relationship exists between the variables. Sadorsky found that that 1 percentage increase in real per capita GDP increases per capita energy consumption by 0.62 percent. A 1 percent increase in real per capita exports increases per capita energy consumption by 0.11 percent while 1 percent increase in real per capita imports increases per capita energy consumption by 0.04 percent. Panel Granger causality analysis revealed that exports Granger cause energy consumption and the feedback is found between imports and energy consumption in short run. Similarly, the bidirectional causality exists between GDP and energy consumption in short run. Sadorsky (2012) used production function to investigate the relationship between trade openness and energy consumption in South American countries namely Argentina, Brazil, Chile, Ecuador, Paraguay, Peru, and Uruguay over the period of 1980-2007. The panel cointegration developed by Pedroni (2004), fully modified ordinary least squares (FMOLS) and the VECM Granger causality approaches were applied. The empirical evidence confirmed the presence of cointegration for long run relationship between the variables. The relationship between exports and energy consumption is bidirectional and imports Granger cause energy consumption in short run. Using data of 52 developed and developing economies, Ghani (2012) explored relationship between trade liberalisation and energy demand. The results indicated that trade liberalisation has insignificant impact on energy consumption but after a certain level of capital per labour, trade liberalisation affects energy consumption.

Hossain (2012) examined the relationship between electricity consumption and exports by adding foreign remittances and economic growth as additional determinants in SAARC countries namely Pakistan, India and Bangladesh. The author reported the no causality between exports and electricity demand. Dedeoglu and Kaya (2013) investigated the relationship between exports, imports and energy consumption by incorporating economic growth as additional determinant of trade openness and energy consumption using data of the OECD countries. They applied the panel cointegration technique developed by Pedroni (2004) and used the Granger causality developed by Canning and Pedroni (2008). Their analysis showed the cointegration between the variables. They also noted that economic growth, exports and imports have positive impact on energy consumption. Their causality analysis revealed that the relationship between exports (imports) and energy consumption is bidirectional.

3. ESTIMATION STRATEGY

Panel Unit Roots

We apply Levine, et al. (2002) (LLC), Im, et al. (2003) (IPS), Maddala and Wu (1999) (MW, ADF) and Maddala and Wu (1999) (MW, PP) panel unit root tests to check the stationarity properties of the variables. These tests apply to a balanced panel but the LLC can be considered a pooled panel unit root test, IPS represents a heterogeneous panel test and MW panel unit root test is non-parametric test.

3.1. LLC Unit Root Test

Levin, et al. (2002) developed a number of pooled panel unit root tests with various specifications depending upon the treatment of the individual specific intercepts and time trends. This test imposes homogeneity on the autoregressive coefficient that indicates the presence or absence of unit root problem while the intercept and the trend can vary across individual series. LLC unit root test follows ADF regression for the investigation of unit root hypothesis as given below step by step:

(1) We use a separate ADF regression for each country:

[DELTA][y.sub.i,t] = [[alpha].sub.i] + [[rho].sub.i] [y.sub.it-1] + [[p.sub.i].summation over (j=1)] [[alpha].sub.i,j] [DELTA][y.sub.i,t-j] + [[epsilon].sub.i,t] ... (1)

The lag order [p.sub.i] is allowable across individual countries. The appropriate lag length is chosen by allowing the maximum lag order and then using the t-statistics for ij b to determine if a smaller lag order is preferred.

(2) We run two separate regressions and save the residuals [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

[DELTA][y.sub.i,t] = [[lambda].sub.i] + [[p.sub.i].summation over (j=1)] [[gamma].sub.i,t-j] [DELTA][y.sub.i,t-j] + [[eta].sub.i,t] [??] [[??].sub.it] ... (2)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)

LLC procedure suggests to standardise the errors [MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] regressing the standard error through the ADF equation provided above:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

(3) Regression can be run to compute the panel test statistics following Equation 5:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (5)

The null hypothesis is as follows: [H.sub.o] : [[rho].sub.1],.. = ... [[rho].sub.n] p = 0 and alternate hypothesis is: [H.sub.A] : [rho] = ... [[rho].sub.n] = [rho] < 0

3.2. IPS Unit Root Test

Im, Pesaran and Shin (IPS), (2003) introduced a panel unit root test in the context of a heterogeneous panel. This test basically applies the ADF test to individual series thus allowing each series to have its own short-run dynamics. But the overall t-test statistic is based on the arithmetic mean of all individual countries' ADF statistic. Suppose a series ([TR.sub.ti,] [EC.sub.ti]) can be represented by the ADF (without trend).

[DELTA][x.sub.i,t] [[bar.[omega]].sub.j] + [[bar.[omega]].sub.i] [x.sub.i,t-1] + [[p.sub.i].summation over (j=1)] [[phi].sub.i,j] [DELTA][x.sub.i,t-j] + [v.sub.i,t] ... (6)

After the ADF regression has different augmentation lags for each country in finite samples, the term E([t.sub.T]) and var([t.sub.T]) are replaced by the corresponding group averages of the tabulated values of E([t.sub.T], [P.sub.i]) and var([t.sub.T],[P.sub.i]) respectively. The IPS test allows for the heterogeneity in the value [[bar.[omega]].sub.i], under the alternative hypothesis. This is more efficient and powerful test than usual single time series test. The estimable equation of IPS unit root test is modeled as follows:

[t.sub.NT] = I/N [N.summation over (i=1)] [t.sub.i,t]([P.sub.i]) ... (7)

where [t.sub.i,t] is the ADF t-statistics for the unit root tests of each country and [P.sub.i] is the lag order in the ADF regression and test statistic can be calculated as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] ... (8)

As [t.sub.NT] is explained above and values for E[[t.sub.iT] ([P.sub.i], 0)] can be obtained from the results of Monte Carlo simulation carried out by IPS. They have calculated and tabulated them for various time periods and lags. When the ADF has different augmentation lags ([P.sub.i]) the two terms E([t.sub.T]) and var([t.sub.T]) in the equation above are replaced by corresponding group averages of the tabulated values of E([t.sub.T], [P.sub.i]) and var([t.sub.T], [P.sub.i]) respectively. (2)

3.3. MW Unit Root Test

The Fisher-type test was developed by Maddala and Wu (1999), which pools the probability values obtained from unit root tests for every cross-section i. This is a non-parametric test and has a chi-square distribution with 2nd degree of freedom where n is number of countries in a panel. The test statistic is given by:

[lambda] = -2 [n.summation over (i=1)] [log.sub.e] ([p.sub.i]) ~ [[chi].sup.2.sub.2n] (d. f.) ... (9)

Where [p.sub.i] is probability value from ADF unit root tests for unit i. The MW unit root test is superior to IPS unit root test because MW unit root test is sensitive to the lag length selection in individual ADF regressions. Maddala and Wu (1999) performed Monte Caro simulations to prove that their test is more advanced than the test developed by IPS (2003).

3.4. The Likelihood-based Panel Cointegration Test

The panel LLL trace test statistics is actually derived from the average of individual likelihood ratio cointegration rank trace test statistics of the panel individuals. The multivariate cointegration trace test of Johanson (1988, 1995) is applied to investigate each individual cross-section system autonomously, in that way, allowing heterogeneity in each cross-sectional unit root for said panel. The process of data generation for each of the groups is characterised by the following heterogeneous VAR ([p.sub.i]) model:

[Y.sub.i,t] = [[p.sub.i].summation over (j=1)] [[LAMBDA].sub.i,j] [Y.sub.i,t-j] + [[epsilon].sub.i,t] ... (10)

Where i = 1, ..., N;t = 1, ... T

For each one, the value of [Y.sub.i,-j+1], ... [Y.sub.i,0] is considered fixed and [[epsilon].sub.i,t] are independent and identically distributed (normally distributed): [epsilon] ~ [N.sub.K] (0, [[OMEGA].sub.i]), where [[OMEGA].sub.i] is the cross-correlation matrix of the error terms: [[OMEGA].sub.i], = E([[epsilon].sub.i,t], [[epsilon].sub.i,t]). The Equation 10 can be modified as vector error correction model (VECM) as given below:

[DELTA][Y.sub.i,t] = [[PI].sub.i][Y.sub.i,t-1] + [[p.sub.i]-1.summation over (j=1)] [[GAMMA].sub.i,j] [DELTA][Y.sub.i,t-j] + [[epsilon].sub.i,j] ... (11)

Where [[PI].sub.i] = [[LAMBDA].sub.i,1] + ... + [[LAMBDA].sub.pi] - 1 and [[GAMMA].sub.i,j] = [[LAMBDA].sub.i,j] - [[GAMMA].sub.i,j-1], [[PI].sub.i], is of order (k x k). If [[PI].sub.i], is of reduced rank: rank ([[PI].sub.i]) = [r.sub.i], which can be de-composed into [[PI].sub.i] = ab, where [[alpha].sub.i] and [[beta].sub.i], are of order (k x [r.sub.i]) and of full column rank that represents the error correction form. The null hypotheses of panel LLL (2001) rank test are:

[H.sub.o] = rank ([[PI].sub.i]) = [r.sub.i] [less than or equal to] r for all i = 1, ..., N against

[H.sub.a] = rank ([[PI].sub.i]) = k for all i = 1, ..., N

The procedure is in sequences like individual trace test process for cointegration rank determination. First, we test for [H.sub.[omicron]] = rank([[PI].sub.i] = [r.sub.i] [less than or equal to] r, r = 0, if null hypothesis of no cointegration is accepted, this shows that there is no cointegration relationship (rank ([[PI].sub.i]) = [r.sub.i] = 0) in all cross-sectional groups for said panel. If null hypothesis is not accepted then null hypothesis r = 1 is tested. The sequence of procedure is not disconnected and continued until null hypothesis is accepted, r = k -1, or is rejected. Accepting the hypothesis of cointegration r = 0 along with null hypothesis of rank ([PI].sub.i]) = r [less than or equal to] 0(0 < r < k) implies that there is at least one cross-sectional unit in panel, which has rank ([[PI].sub.i]) = r > 0. The likelihood ratio trace test statistic for group i is as following;

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] ... (12)

Where [[lambda]'.sub.l] is the lth largest eigen value in the ith cross-section unit. The LR-bar statistic is calculated as the average of individual trace statistics:

L [bar.R]iT [H(r)/H(k)] = 1/N [n.summation over (i=1)] [LR.sub.iT] [H(r)/H(k)] ... (13)

Finally, modified version of above equation is defined as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (14)

Where E([Z.sub.k]) and Var([Z.sub.k]) are mean and variance of the asymptotic trace statistics, which can be obtained from simulation. The LLL (2001) proves the central limit theorem for the standard LR-bar statistic, according to which under the null hypothesis, [[lambda].sub.LR] [??] N(0,1) as N and T [right arrow] [infinity] in such a way that [[square root of NT].sup.-1] [right arrow] 0, under the assumption that there is no cross-correlation in the error terms, that is given below:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

LLL (2001) notes that T [right arrow] [infinity] is needed for each of the individual test statistic to converge to its asymptotic distribution, while N [right arrow] [infinity] is needed for the central limit theorem.

3.5. Panel Causality Test

Hurlin and Venet (2001) extended the Granger (1969) causality test for panel data models with fixed coefficients. The estimable equation for empirical estimation is modeled as following:

[y.sub.i,t] = [P.summation over (K=1)] [[gamma].sup.(K)] [y.sub.i,t-K] + [P.summation over (K=0)] [[beta].sup.(K).sub.i] [x.sub.i,t-K] + [v.sub.i,t] ... (15)

With P [member of] [N.sup.*] and [v.sub.i,t] = [[partial derivative].sub.i] + [[epsilon].sub.i,t], where [[epsilon].sub.i,t], are i.i.d (O, [[sigma].sup.2][epsilon]). In contrast to Nair-Reichert and Weinhold (2001), we assume that the autoregressive coefficients [[gamma].sup.(k) and the regression coefficients slopes [[beta].sub.i.sup.(k)] are constant [OMEGA]k[epsilon][1, p]. We also assume that parameters [[gamma].sup.(k)] are identical for all individuals, whereas the regression coefficients slopes [[beta].sup.(K).sub.i] could have an individual dimension. Hurlin and Venet (2001), consider four principal cases following Equation 15.

3.6. Homogenous Non-Causality Test

Initially the homogenous non-causality (HNC) hypothesis has been discussed. Conditional to the specific error components of the model, this hypothesis assumes no prevalence of any individual causality association:

[for all]i [member of] [1, N] E ([y.sub.i,t]/[[bar.y].sub.i,t],[[alpha].sub.i]) = E ([y.sub.i,t]/[[bar.y].sub.i,t],[[bar.x].sub.i,t],[[alpha].sub.i]) ... (16)

In Equation 15, the corresponding test (3) is defined by:

[H.sub.o]: [[beta].sup.(K).sub.i] = 0 [[for all].sub.i] [member of] [1,N], [for all]k [member of] (1, p) ... (17)

[H.sub.a] : [there exists](i, k)/[[beta].sup.(K).sub.i] [not equal to] 0

In order to test these [N.sub.p] linear restrictions Wald Statistic is employed:

[F.sub.hnc] = ([RSS.sub.2] - [RSS.sub.1])/(Np)/[RSS.sub.1]/[NT - N(1 + p) - p] ... (18)

Where [RSS.sub.2] indicates the restricted sum of squared residuals. [RSS.sub.1] corresponds to the residual sum of squares of equation-15. If the realisation of this statistic is not significant, the homogeneous non-causality hypothesis is accepted. This result implies that the variable X is not causing Y in finite sample set in all countries. If the non-causality result is totally homogenous then further empirical exercise is stopped.

3.7. Homogenous Causality Test

Secondly, homogenous causality (HC) hypothesis is proven, in which there exist N causality relationships:

[for all]i [member of] [1, N] E ([y.sub.i,t]/[[bar.y].sub.i,t],[[alpha].sub.i]) [not equal to] E ([y.sub.i,t]/[[bar.y].sub.i,t],[[bar.x].sub.i,t], [[alpha].sub.i]) ... (19)

In this case, suppose that the N individual predictors, obtained conditional to the fact that [[bar.Y].sub.i,t], [[bar.X].sub.i,t] and [[alpha].sub.i], are the same:

[for all] (i,j) [member of] [1, N] E ([y.sub.i,t]/[[bar.y].sub.i,t],[[bar.x].sub.i,t] [[alpha].sub.i]) = E ([y.sub.i,t]/[[bar.y].sub.j,t],[[bar.x].sub.j,t],[[alpha].sub.j]) ... (20)

Two configurations could appear, if we reject hypothesis of non-homogenous causality. The first one corresponds to the overall causality hypothesis (homogenous causality hypothesis) and occurs if all the coefficients [[beta].sup.K.sub.i] are identical for all k. The second one is more plausible, which is that some coefficients [[beta].sup.K.sub.i] are different for each individual. Thus, after the rejection of the null hypothesis of non-homogenous causality, the second step of the procedure consists of testing if the regression slope coefficients associated [x.sub.i,t-k] are identical. This test corresponds to a standard homogeneity test. Formally, the homogenous causality hypothesis test is as following:

[H.sub.o]: [for all]k [member of] [1,p]/[[beta].sup.k.sub.i] = [[beta].sup.k] [for all]i [member of] [1,N] ... (21)

[H.sub.a]: [their exists]k [member of] [1,p], [there exists] (i,j) [member of] [1,N]/[[beta].sup.k.sub.i] [not equal to] [[beta].sup.k.sub.j]

The homogenous causality hypothesis implies that the coefficients of the lagged explanatory variables [x.sup.i,t-k] are identical for each lag k and different from zero. Indeed, if we have rejected, in the previous step, the non-homogenous causality hypothesis [[beta].sup.K.sub.i] = 0 [for all](i,k), this standard specification test allows testing the homogenous causality hypothesis. In order to test the homogenous causality hypothesis, F-statistic is calculated by applying the given mechanism:

[F.sub.hc] = ([RSS.sub.3] - [RSS.sub.1])/[p(N - 1)/[RSS.sub.1]/[NT - N(1 + p) - p] ... (22)

where, [RSS.sub.3] corresponds to the realisation of the residual sum of squares obtained in Equation 15 when one imposes the homogeneity for each lag k of the coefficients associated to the variable [x.sub.i,t-k]. If the [F.sub.hc] statistics with P(N -1) and NT - N(1 + P) - P degrees of freedom is not significant, the homogenous causality hypothesis is accepted. This result implies that the variable X is causing Y in the N countries of the samples, and that the autoregressive processes are completely homogenous.

3.8. Heterogeneous Causality Test

Third case is relevant to the heterogeneous causality hypothesis. Under HEC hypothesis, it is assumed there exists at least one individual causality relationship (and at the most N), and second that individual predictors, obtained conditional to the fact that [[bar.y].sub.i,t], [[bar.x].sub.i,t], [[bar.[lambda].sub.t] and, [[alpha].sub.i] are heterogeneous.

[there exists]i [member of] [1,N] E([y.sub.i,t]/[[bar.y].sub.i,t], [[alpha].sub.i]) [not equal to] E([y.sub.i,t]/[[bar.y].sub.i,t], [[bar.x].sub.i,t], [[alpha].sub.i]) ... (23)

[there exists](i,j) [member of] [1,N] E([y.sub.i,t]/[[bar.y].sub.i,t], [[bar.x].sub.i,t], [[alpha].sub.i]) [not equal to] E([y.sub.j,t]/[[bar.y].sub.j,t], [[bar.x].sub.j,t], [[alpha].sub.j]) ... (24)

3.9. Heterogeneous Non-causality Test

Finally, heterogeneous non-causality hypothesis assumes that there exists at least one and at the most N-1 equalities of the form:

[there exists]i [member of] [1,N] E([y.sub.i,t]/[[bar.y].sub.i,t], [[alpha].sub.i]) = E([y.sub.i,t]/[[bar.y].sub.i,t], [[bar.x].sub.i,t], [[alpha].sub.i]) ... (25)

The third step of the procedure consists of testing the heterogeneous non-causality hypothesis (HENC). The following equation explains this mechanism:

[H.sub.o]: [there exists]i [member of] [1,N]/[for all]k [member of] [1, p] [[beta].sup.K.sub.i] = 0 ... (26)

[H.sub.a]: [for all]i [epsilon] [1,N]/[there exists]k [epsilon] [1, N]/[[beta].sup.K.sub.i] [not equal to] 0

This test is proposed to test this last hypothesis with two nested tests. The first test is an individual test realised for each individual. For each individual i = 1 ... N, we test the nullity of all the coefficients of the lagged explanatory variables [x.sub.i,t-k]. Then, for each i, we test the hypothesis [[beta].sup.K.sub.i] = 0, [for all]k [member of] [1,p]. For that, we compute N statistics:

[F.sup.i.sub.hene] = ([RSS.sub.2,i] - [RSS.sub.1])/p/[RSS.sub.1]/[NT - N(1 + 2p) + p] ... (27)

where, [RSS.sub.2,i] corresponds to the realisation of the residual sum of squares obtained in model (15), when one imposes the nullity of the k coefficients associated to the variable [x.sub.i,t-k] only for the individual i. A second test of the procedure consists of testing the joint hypothesis that there is no causality relationship for a sub-group of individuals. Let us respectively denote [I.sub.c] and [I.sub.nc] as the index sets corresponding to sub-groups for which there exists a causal relationship and there does not exist a causal relationship. In other words, we consider the following model [for all]t [member of] [1, T]:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (28)

with

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII]

Let [n.sub.c] = dim([I.sub.c]) and [n.sub.nc]=dim ([I.sub.nc]). Suppose that [n.sub.c]/[n.sub.nc] [right arrow] [right arrow] [theta]< [infinity] as [n.sub.c] and [n.sub.nc] tend to infinity. One solution to test the HENC hypothesis is to compute the Wald statistic.

[F.sub.henc] = ([RSS.sub.4] - [RSS.sub.1])/([n.sub.nc]p)/[RSS.sub.1][NT - N(1 + p) - [n.sub.c]p] ... (29)

where [RSS.sub.4] corresponds to realisation of the residual sum of squares obtained from equation-15 when one imposes the nullity of the k coefficients associated to the variable [x.sub.i,t-k] for the [n.sub.nc] individuals of the [I.sub.nc] sub-group. If the HENC hypothesis is accepted, it implies that there exists a sub-group of individuals for which the variable x does not cause the variable y. The dimension of this sub-group is then equal to [n.sub.nc]. On the contrary, if the HENC hypothesis is rejected, it implies that there exists a causality between x and y for all individuals of the panel.

3.10. Data and Data Sources

The 91 countries are selected for the estimation of causality between energy consumption and trade openness on the basis of data availability. (4) The study covers the period 1980-2010. All necessary data for the sample period are obtained from World development Indicators (CD- ROM, 2012). Energy consumption in kg of oil equivalent per capita is used to measure energy consumption, real exports (US$) plus real imports (US$) divided by population are used to measure trade openness. Both variables are used in their natural logarithmic form.

4. EMPIRICAL RESULTS AND THEIR DISCUSSIONS

The results of ADF unit root test in the presence of intercept and, intercept and trend reported in Table 1 suggest that all the series are non-stationary at their level, but stationary at first difference. This implies that real trade per capita ([TR.sub.t]) and energy consumption per capita ([EC.sub.t]) are integrated at I(1) for each country in our sample.

The unit root test results set the stage for Johansen cointegration approach. The results are presented in Table 2. We find the acceptance of null hypothesis i.e. no cointegration in case of Angola, Brazil, Bulgaria, Cameroon, Congo Dem Rep, Congo Rep, Israel, Italy, Kenya, South Korea, Kuwait, Nicaragua, Pakistan, Panama, Philippines, Sudan, Tunisia, Turkey, Zambia and Zimbabwe. We find two cointegrating vectors in case of Benin, Saudi Arabia, Cyprus, Denmark, Ecuador, Ghana, Indonesia, Luxemburg and Paraguay and for the rest of countries, we find one cointegrating vector. The existence of one or two cointegrating vectors confirms the presence of cointegration between the variables. This shows that trade openness and energy consumption have long run relationship over selected period of time i.e., 1980-2010.

This ambiguity in the results based on single country study prompts us to apply panel cointegration approach. (5) For this purpose, we apply panel unit root tests to check for stationary properties of the series. The results based on the LLC, IPS, MW (ADF) and MW (PP) unit root tests with constant and, constant and trend are reported in Table 3. The tests show that all variables are found to be non-stationary at level. At first difference, all the series are integrated i.e. I(1). This unique order of integration of the variables helps us to apply Johansen panel cointegration approach to examine long run relationship between the variables for selected panel.

The results are reported in Table 4. We find that maximum likelihood ratio i.e. 5.9035 is greater than critical value at 1 percent level of significance. This leads us to reject the null hypothesis of no panel cointegration between the variables. We may conclude that the panel cointegration exists between trade openness and energy consumption in sampled countries. The Table 5 shows that trade openness affects energy consumption in high, middle and low-income countries. In high-income countries, we find that the relationship between trade openness and energy consumption is inverted U-shaped. This implies that initially trade openness is positively linked with energy consumption and after a threshold level, it declines energy demand due to adoption of energy efficient technology. This indicates that a 1 percent increase in trade openness raises energy demand by 0.860 percent and negative sign of nonlinear term of trade openness corroborates the delinking of energy consumption as trade openness is at optimal level. In case of middle and low income countries, relationship between trade openness and energy consumption is U-shaped which reveals that trade openness decreases energy consumption initially but energy consumption is increased with continuous process of trade openness. In middle-income countries, trade openness stimulates industrialisation, which raises energy demand [Cole (2006)]. It is argued by Ghani (2006) that low-income countries are unable to reap optimal fruits of trade liberalisation because these economies are lacking in utilisation of energy efficient technology to enhance domestic production.

The presence of cointegration between the series leads us to investigate the direction of causality. In doing so, we have applied homogeneous and non-homogenous panel causality and results are reported in Table 6. The results of non-homogenous causality reveal the feedback hypothesis between trade openness and energy consumption as bidirectional causal relationship is confirmed between both the series. We find that trade openness Granger causes energy consumption confirmed by homogeneous causality (see Table 6).

Our results of non-homogenous causality validated the presence of feedback effect, as trade openness and energy consumption are interdependent. The unidirectional causality is found running from trade openness to energy consumption. This validates the presence of trade-led-energy hypothesis confirmed by homogenous causality approach. This ambiguity in results would not be helpful in policy-making and leads us to apply homogenous and non-homogenous causality approach using data of low, middle and high-income countries. This will not only help us in obtaining results region-wise but also enable us to design a comprehensive trade and energy policy for sustained economic growth and better living standards. In doing so, we have investigated the homogenous and non-homogenous causal relationship separately for high, middle and low-income countries. The results are reports in Table 7. In high income countries, non-homogenous causality confirms the unidirectional causality running from trade openness to energy consumption but feedback effect is confirmed by homogenous causality between both variables. The relationship between trade openness and energy consumption is bidirectional for middle and low-income countries confirmed by homogenous and non-homogenous causality approaches.

The results of heterogeneous causality reported in Table 7 suggest the feedback relationship between trade openness and energy consumption i.e. bidirectional causality exists in case of Albania, Cote D'Ivoire, Cyprus, Egypt, Finland, Gabon, Honduras, Hong Kong, Kuwait, Morocco, Norway, Panama, Saudi Arabia, Togo, Uruguay and Unites States. Energy consumption Granger causes trade openness in case of Bangladesh, Benin, China, Cuba, Ecuador, Ethiopia, France, Greece, Guatemala, Indonesia, Mozambique, The Netherlands, Nicaragua, Paraguay, Philippines, Sweden, Switzerland, Trinidad and Tobago, United Arab Emirates, Venezuela and Zambia.

The unidirectional causality is found running from trade openness to energy consumption. This validates the trade-led-energy hypothesis in case of Algeria, Angola, Argentina, Australia, Botswana, Brazil, Chili, Costa Rica, El Salvador, Ghana, India, Italy, Japan, Jordan, Mexico, Nepal, Oman, Pakistan, Portugal, Sudan, Thailand and Turkey. The neutral relationship between trade openness and energy consumption i.e. no causality exists between both the variables for Austria, Belgium, Bolivia, Brunei Darussalam, Bulgaria, Cameroon, Canada, Colombia, Congo Democratic Republic, Congo Republic, Denmark, Dominican Republic, Hungary, Iceland, Iran, Ireland, Israel, Kenya, South Korea, Luxemburg, New Zealand, Nigeria, Senegal, South Africa, Spain, Syria, Tunisia, United Kingdom, Vietnam and Zimbabwe.

5. CONCLUDING REMARKS AND FUTURE DIRECTIONS

This paper explores the relationship between trade openness and energy consumption using data of 91 heterogeneous (high, middle and low income) countries over the period of 1980-2010. In doing so, we have applied time series as well as panel unit root tests to examine the integrating properties of the variables. Similarly, to examine cointegration between the variables, we have applied single country as well as panel cointegration approaches. The homogenous and non-homogenous causality approaches are applied to examine the direction of causality between the variables in high, middle and low-income countries. Heterogeneous causality approach has also been applied to examine relationship between trade openness and energy consumption at country level analysis.

Our results indicated that our variables are integrated at I(1) confirmed by time series and panel unit root tests and same is inference is drawn about cointegration between trade openness and energy consumption. The pooled mean group estimation analysis reveals an inverted U-shaped relationship in high income countries and vice versa in middle and low income countries. The causality analysis confirms the existence of feedback effect between trade openness and energy consumption in middle and low income countries but bidirectional causality is confirmed by homogenous causality approach in high income countries, however non-homogenous causality approach indicates unidirectional causality running form trade openness to energy consumption. Heterogeneous causality exposes that in 18 percent of sampled countries, the feedback effect exists while 24 percent show that trade openness causes energy consumption and rest of sample countries confirm the presence of neutral relationship between trade openness and energy consumption.

Overall, our results demonstrate that the feedback effect exists between trade openness and energy consumption, which suggests in exploring new and alternative sources of energy to reap optimal fruits of trade. Trade openness stimulates industrialisation that in turn affects economic growth. This channel of trade affects energy demand via economic growth. Similarly, insufficient energy supply impedes economic growth, which affects exports as well as imports, and as a result energy consumption decreases. Trade openness also is a source of transferring advanced technologies i.e. energy efficient technology from developed countries to developing economies. Our findings confirm that the relationship between trade openness and energy consumption is U-shaped. This suggests that middle and low-income countries should import energy efficient technologies from developed economies to lower energy intensity. This will not be possible if developed countries do not promote those technologies and lower prices for countries, which do not have access to required amounts of capital. Further, it will have a positive impact on the world economy as it will save natural resources for future generations and it will reduce environmental pollution.

This paper can be augmented for future research by incorporating financial development, industrialisation, urbanisation in energy demand function following Shahbaz and Lean (2012) in case of low, middle and high-income countries. The semi-parametric panel approach proposed by Baltagi and Lu (2002) could be applied to investigate the impact of financial development, industrialisation, trade openness and urbanisation on energy consumption using global level data. Using global level data, trade openness, financial development, industrialisation, urbanisation and CO2 emissions nexus could be investigated by applying heterogamous panel under cross-sectional dependence framework.

Muhammad Shahbaz <shahbazmohd@live.com> is affiliated with the COMSATS Institute of Information Technology, Lahore. Sarnia Nasreen <sami_leu@yahoo.com> is affiliated with the Government College University Faisalabad, Faisalabad Chong Hui Ling <felicia.chonghl@gmail.com> is affiliated with the University of Malaya Kuala Lumpur Malaysia. Rashid Sbia <Rashid.Sbia@ulb.ac.be> is affiliated with the Department of Applied Economics, Free University of Brussels, Brussels, Belgium.

APPENDIX A
List of World Countries

High Income Countries   Middle Income Countries   Low Income Countries

Angola                  Algeria                   Bangladesh
Australia               Argentina                 Benin
Austria                 Bolivia                   Congo Dem Rep
Albania                 Botswana                  Ethiopia
Belgium                 Brazil                    Kenya
Brunei Darussalam       Bulgaria                  Mozambique
Canada                  Cameroon                  Nepal
Cyprus                  Chile                     Togo
Denmark                 China                     Zimbabwe
Finland                 Colombia
France                  Congo Rep
Greece                  Costa Rica
Hong Kong               Cote D'Ivoire
Hungary                 Cuba
Iceland                 Dominican Rep
Israel                  Ecuador
Italy                   Egypt
South Korea             El Salvador
Kuwait                  Gabon
Luxemburg               Ghana
The Netherlands         Guatemala
New Zealand             Honduras
Norway                  India
Oman                    Indonesia
Portugal                Iran
Saudi Arabia            Ireland
Spain                   Jamaica
Sweden                  Japan
Switzerland             Jordan
Trinidad and Tobago     Mexico
United Kingdom          Morocco
United Arab Emirates    Nicaragua
Unites States           Nigeria
                        Pakistan
                        Panama
                        Paraguay
                        Peru
                        Philippines
                        Senegal
                        South Africa
                        Sudan
                        Syria
                        Thailand
                        Tunisia
                        Turkey
                        Uruguay
                        Venezuela
                        Vietnam
                        Zambia

REFERENCES

Akarca, A. T. and T. V. Long (1979) Energy and Employment: A Time Series Analysis of The Causal Relationship. Resources and Energy 5, 326-331.

Akarca, A. T. and T. V. Long (1980). On the Relationship between Energy and GNP: A Re-examination. Journal of Energy and Development 5, 326-331.

Amirat, A. and A. Bouri (2010) Energy and Economic Growth: The Algerian Case.

Apergis, N. and E. J. Payne (2010) Energy Consumption and Growth in South America: Evidence from a Panel Error Correction Model. Energy Economics 32, 1421-1426.

Arrow, K. J. (1962) The Economic Implications of Learning by Doing. The Review of Economic Studies 29, 155-173.

Baltagi, B. H. and D. Li (2002) Series Estimation of Partially Linear Panel Data Models with Fixed Effects. Annals of Economics and Finance 3, 103-116.

Canning, D. and P. Pedroni (2008) Infrastructure, Long-run Economic Growth and Causality Tests for Cointegrated Panels. Manchester School, 76, 504-527.

Cole, M. A. (2006) Does Trade Liberalisation Increase Energy Use? Economics Letters 92, 108-112.

Dedeoglu, D. and H. Kaya (2013) Energy Use, Exports, Imports and GDP: New Evidence from the OECD Countries. Energy Policy 57, 469-476.

Ghani, G. M. (2012) Does Trade Liberalisation Affect Energy Consumption? Energy Policy 43:4, 285-290.

Granger, C. W. J. (1969) Investigating Causal Relations by Econometric Models and Cross-Spectral Methods. Econometrica 37, 424-438.

Halicioglu, F. (2011) A Dynamic Study of Income, Energy and Exports in Turkey. Energy 36, 3348-3354.

Hossain, Md. S. (2012) Multivariate Granger Causality between Economic Growth, Electricity Consumption, Exports and Remittances for the Panel of three SAARC Countries. Global Journal of Management and Business Research 12, 41-54.

Hurlin, C. and B. Venet (2001) Granger Causality Tests in Panel Data Models with fixed Coefficients, 12th (EC)2 Conference on Causality and Exogeneity in Econometrics, Core Louvain--Ia--Neuve.

Im, K. S., M. H. Pesaran, and Y. Shin (2003) Testing Unit Roots in Heterogeneous Panels. Journal of Econometrics 115, 53-74.

Jena, P. R. and U. Grote (2008) Growth-trade-environment Nexus in India. Economics Bulletin 17, 1-17.

Johansen, S. (1988) Statistical Analysis of Cointegration Vectors. Journal of Economic Dynamics and Control 12, 231-254.

Johansen, S. (1995) A Statistical Analysis of Cointegration for I(2) Variables. Econometric Theory 11, 25-59.

Karlsson, S. and M. Lothgren (2000) On the Power and Interpretation of Panel Unit Root Tests. Economics Letters 66, 249-255.

Kraft, J. and A. Kraft (1978) On the Relationship between Energy and GNP. Journal of Energy and Development 3, 401-403.

Larsson, R., J. Lyahagen, and M. Lothgren (2001) Likelihood-based Cointegration Tests in Heterogeneous Panels. Econometrics Journal 4, 109-142.

Lean, H. H. and R. Smyth (2010) On the Dynamics of Aggregate Output, Electricity Consumption and Exports in Malaysia: Evidence from Multivariate Granger Causality Tests. Applied Energy 87, 1963-1971.

Levin, A., C-F. Lin, and C-S. J. Chu (2002) Unit Root Tests in Panel Data: Asymptotic and Finite-sample Properties. Journal of Econometrics 108, 1-24.

Maddala, G. S. and S. Wu (1999) A Comparative Study of Unit Root Tests with Panel Data and New Simple Test. Oxford Bulletin of Economics and Statistics 631-652.

Masih, A. M. M. and R. Masih (1996) Energy Consumption, Real Income and Temporal Causality Results from A Multi-Country Study Based on Cointegration and Error Correction Modeling Techniques. Energy Economics 18, 165-183.

Nair-Reichert, U. and D. Weinhold (2001) Causality Tests for Cross-Country Panels: A New Look at FDI and Economic Growth in Developing Countries. Oxford Bulletin of Economics and Statistics 63, 153-171.

Narayan, P. K., S. Narayan, and A. Prasad (2008) A Structural VAR Analysis of Electricity Consumption and Real GDP: Evidence from the G7 Countries. Energy Policy 36, 2765-2769.

Narayan, P. K. and R. Smyth (2009) Multivariate Granger Causality between Electricity Consumption, Exports and GDP: Evidence from a Panel of Middle Eastern Countries. Energy Policy 37, 229-236.

Nondo, C. and M. Kahsai (2009) Energy Consumption and Economic Growth: Evidence from COMESA Countries. Selected Paper prepared for presentation at the Southern Agricultural Economics Association Annual Meeting, Atlanta, Georgia, January 31-February 3.

Noor, S. and M. W. Siddiqi (2010) Energy Consumption and Economic Growth in South Asian Countries: A Cointegrated Panel Analysis. International Journal of Human and Social Sciences 5, 921-926.

Ozturk, I. (2010) A Literature Survey on Energy-Growth Nexus. Energy Policy 38:1, 340-349.

Payne, J. (2010) A Survey of the Electricity Consumption-Growth Literature. Applied Energy 87, 3723-3731.

Pedroni, P. (1999) Critical Values for Cointegration Tests in Heterogeneous Panels with Multiple Regressors. Oxford Bulletin of Economics and Statistics, Special Issue 61, 653-670.

Pedroni, P. (2004) Panel Cointegration: Asymptotic and Finite Samples Properties of Pooled Time Series Tests with an Application to the PPP Hypothesis. Econometric Theory 2D, 597-625.

Sadorsky, P. (2011) Trade and Energy Consumption in the Middle East. Energy Economics 33, 739-749.

Sadorsky, P. (2012) Energy Consumption, Output and Trade in South America. Energy Economics 34, 476-488.

Shahbaz, M. and H. H. Lean (2012) Does Financial Development Increase Energy Consumption? The Role of Industrialisation and Urbanisation in' Tunisia. Energy Policy 40, 473-479.

Shahbaz, M., S. Khan, and M. I. Tahir (2013b) The Dynamic Links between Energy Consumption, Economic Growth, Financial Development and Trade in China: Fresh Evidence from Multivariate Framework Analysis. Energy Economics 40, 8-21.

Shahbaz, M., H. H. Lean, and A. Farooq (2013a) Natural Gas Consumption and Economic Growth in Pakistan. Renewable and Sustainable Energy Reviews 18, 87-94.

Yang, H. Y. (2000) A note on the Causal Relationship between Energy and GDP in Taiwan. Energy Economics 22, 309-317.

Yu, E. S. H. and J. Y. Choi (1985) The Causal Relationship between Energy and GNP: An International Comparison. Journal of Energy and Development 10:2, 249-272.

Zahid, K. (2008) Energy-GDP Relationship: A Casual Analysis for the Five Countries of South Asia. Applied Econometrics and International Development 8, 167-180.

Comments

Paper gives a good comparison among the high, middle and low income countries in terms of energy usage. Few comments which can improve the paper are; inclusion of the role of mediating/moderating variable which is production through which energy has causal relationship between trade openness. Baron and Kenny (1986) (1) gives a good technique of using moderating/mediating variable. Battery of tests is estimations are done in the paper but authors are very miser to explain the results. Since the panel data estimation is done to obtain the estimates therefore there is no need for single country regression or if authors have different objective in their mind then they did not explain it in the text. The paper says that 25 percent of the sample countries have positive association between energy and trade openness, what would author infer from this result. Since the data is from 1980-2010, thus I would recommend to apply a structural break test on the variables.

M. Ali Kemal

Pakistan Institute of Development Economics, Islamabad.

(1) Reuben M. Baron and David A. Kenny (1986) "The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations", Journal of Personality and Social Psychology, Vol. 51, No. 6, 1173-1182.

(1) Payne (2010) and Ozturk (2010) presented comprehensive survey studies on the relationship between economic growth and energy consumption.

(2) Karlsson and Lothgren (2000) demonstrate the power of panel unit root tests by Monte Carlo simulation. The null of all these tests is that each series contains a unit root and thus is difference stationary. However, the alternative hypothesis is not clearly specified. In LLC the alternative hypothesis is that all individual series in the panel are stationary. In IPS the alternative hypothesis is that at least one of the individual series in the panel is stationary. They conclude that the "presence or absence of power against the alternative hypothesis where a subset of the series is stationary has a serious implications for empirical work. If the tests have high power, a rejection of the unit root null can be driven by few stationary series and the whole panel may inaccurately be modelled as stationary. If, on the other hand, the tests have low power it may incorrectly concluded that the panel contains a common unit root even if a majority of the series is stationary" (p. 254). The simulation results reveal that the power of the tests (LLC, IPS) increases monotonically with: (1) an increased number (N) of the series in the panel; (2) an increased time series dimension (T) in each individual series; (3) increased proportion of stationary series in the panel. Their Monte Carlo simulations for N=13 and T=80 reveal the power of the test is 0.7 for LLC tests and approaching unity for the IPS tests.

(3) Here, we do not consider instantaneous non-causality hypothesis.

(4) The selection of countries is restricted to availability of data. The names of countries are listed in Appendix-A.

(5) In some countries we could not find cointegration while in rest of the countries we found the existence of cointegration between the variables.

(6) Hausman test indicates that PMG model is preferred over PG model.

(7) A graph is provided in Appendix for high income countries.

Table 1
ADF Unit Root Test

                           Level                    1st Difference

Country/          Intercept    Trend &      Intercept       Trend &
Variable                      Intercept                    Intercept

Algeria
[TR.sub.t]         0.4189     -0.8701     -3.8052 **      -5.1733 *
[EC.sub.t]        -0.6407     -1.4528     -5.8948 *       -5.2814 *
Argentina
[TR.sub.t]        -1.0531     -3.0792     -5.2571 *       -5.0271 *
[EC.sub.t]        -0.8932     -2.8109     -3.6245 **      -3.6308 **
Austria
[TR.sub.t]        -0.5524     -2.4505     -3.2985 **      -3.5066 ***
[EC.sub.t]        -0.1863     -2.5139     -4.6619 *       -4.4885 *
Bangladesh
[TR.sub.t]         0.6132     -3.0994     -3.9199 *       -3.9065 **
[EC.sub.t]         1.0205     -2.3929     -4.6232 *       -5.1651 *
Benin
[TR.sub.t]        -0.3299     -2.3450     -4.9286 *       -5.0471 *
[EC.sub.t]        -1.9601     -2.6871     -3.5797 **      -3.5434 ***
Botswana
[TR.sub.t]        -1.4420     -2.4192     -3.9853 *       -4.0636 **
[EC.sub.t]        -1.0734     -1.3623     -3.0628 **      -5.6302 *
Brunei Darussalam
[TR.sub.t]        -0.3508     -1.4825     -3.6958 **      -5.7109 *
[EC.sub.t]        -1.9429     -3.1187     -3.7129 **      -3.6122 ***
Canada
[TR.sub.t]        -1.9408     -2.4400     -4.9088 *       -5.2583 *
[EC.sub.t]        -2.0028     -3.1663     -3.7820 *       -3.7348 **
Chili
[TR.sub.t]        -0.7908     -2.4845     -5.5118 *       -5.3639 *
[EC.sub.t]         0.3533     -2.8041     -2.9216 ***     -4.6043 *
Angola
[TR.sub.t]         1.5123     -0.5634     -3.5182 **      -4.5661 *
[EC.sub.t]        -1.6214     -1.5625     -3.2417 **      -5.9735 *
Australia
[TR.sub.t]         0.3937     -2.6913     -4.3756 *       -4.5020 *
[EC.sub.t]         0.1996     -2.7783     -4.1198 *       -4.2963 **
Albania
[TR.sub.t]        -0.7642     -1.6930     -4.4905 *       -4.9971 *
[EC.sub.t]        -1.5043     -1.2434     -3.0995 **      -3.2659 ***
Belgium
[TR.sub.t]        -0.5282     -2.2922     -3.0316 **      -3.5863 ***
[EC.sub.t]        -1.9601     -2.6871     -3.5797 **      -3.5434 ***
Bolivia
[TR.sub.t]         0.2859     -1.3079     -2.9710 ***     -4.3259 **
[EC.sub.t]        -1.4582     -2.1065     -3.5069 **      -3.4382 ***
Brazil
[TR.sub.t]         1.1870     -2.1045     -4.5757 *       -4.8461 *
[EC.sub.t]        -0.9027     -2.4494     -3.1364 **      -3.7495 **
Bulgaria
[TR.sub.t]        -0.4585     -0.4585     -2.7263 ***     -4.3906 **
[EC.sub.t]        -1.3805     -2.2254     -3.3030 **      -3.9770 **
China
[TR.sub.t]         0.1074     -2.1102     -4.8452 *       -4.8994 *
[EC.sub.t]         0.6452     -2.0721     -2.9494 **      -3.2235 ***
Congo Dem Rep
[TR.sub.t]        -2.5579     -2.8169     -3.9579 *       -3.8466 **
[EC.sub.t]        -0.6483     -1.9564     -4.2579 *       -4.1745 **
Colombia
[TR.sub.t]        -0.0635     -2.6416     -3.1969 **      -1.5686 *
[EC.sub.t]        -1.1615     -1.4324     -4.8072 *       -4.8553 *
Congo Rep
[TR.sub.t]        -1.5302     -2.7516     -3.9847 *       -3.8813 **
[EC.sub.t]        -1.2094     -0.5212     -3.2900 **      -3.4620 ***
Cote D'Ivoire
[TR.sub.t]         0.2225     -1.9929     -3.6169 **      -3.8302 **
[EC.sub.t]        -0.9567     -1.7444     -3.9964 *       -4.8263 *
Cuba
[TR.sub.t]        -1.8938     -1.6057     -2.7562 ***     -3.9406 **
[EC.sub.t]        -1.4306     -2.8859     -2.9979 **      -2.9527 ***
Denmark
[TR.sub.t]        -0.0910     -2.3117     -3.2089 **      -3.5203 ***
[EC.sub.t]        -2.0518     -2.7916     -3.7190 **      -3.6570 **
Ecuador
[TR.sub.t]         0.7030     -2.0413     -3.4003 **      -3.9494 **
[EC.sub.t]        -0.1665     -1.1361     -3.3996 **      -4.2587 **
El Salvador
[TR.sub.t]        -0.0745     -2.2870     -3.4843 **      -3.3700 ***
[EC.sub.t]        -0.0416     -1.7824     -2.8539 ***     -3.7315 **
Finland
[TR.sub.t]        -0.6923     -2.7347     -3.7078 **      -3.5774 ***
[EC.sub.t]        -2.3395     -2.7686     -4.3644 *       -4.1951 **
Gabon
[TR.sub.t]        -0.9361     -2.7341     -3.9640 *       -4.2463 **
[EC.sub.t]        -2.2723     -1.0959     -3.5525 **      -4.5870 *
Greece
[TR.sub.t]         0.5889     -2.8057     -3.5020 **      -3.6567 **
[EC.sub.t]        -1.8250     -2.0913     -4.5134 *       -5.0303 *
Costa Rica
[TR.sub.t]        -0.2737     -2.3264     -3.6127 **      -3.5250 ***
[EC.sub.t]        -0.2865     -0.3390     -3.2568 **      -3.8902 **
Cameroon
[TR.sub.t]        -1.5618     -2.9541     -2.7506 ***     -5.6762 *
[EC.sub.t]        -1.0496     -1.0088     -3.6118 **      -4.1561 **
Cyprus
[TR.sub.t]        -0.4131     -1.6628     -3.3912 **      -3.3175 ***
[EC.sub.t]        -1.5058     -0.5346     -3.3796 **      -3.8715 **
Dominican Rep
[TR.sub.t]        -0.5985     -2.1949     -5.3140 *       -5.2511 *
[EC.sub.t]        -0.9124     -1.6794     -3.9453 *       -3.8494 **
Egypt
[TR.sub.t]         0.5745     -2.7622     -2.7713 ***     -3.6586 **
[EC.sub.t]        -1.0024     -2.4033     -3.5517 **      -3.3564 ***
Ethiopia
[TR.sub.t]        -0.0839     -1.2336     -4.3298 *       -4.6814 *
[EC.sub.t]        -1.4764     -1.9549     -3.2659 **      -3.8596 **
France
[TR.sub.t]        -0.4312     -2.3780     -3.2569 **      -3.6901 **
[EC.sub.t]        -1.3933     -1.7466     -4.2313 *       -4.6509 *
Ghana
[TR.sub.t]        -1.7857     -1.5640     -5.0802 *       -5.4612 *
[EC.sub.t]        -1.0468     -1.0777     -4.1390 *       -4.2675 **
Guatemala
[TR.sub.t]         0.7712     -3.0441     -3.3703 **      -3.6195 **
[EC.sub.t]        -1.3829     -2.0519     -3.3144 **      -3.4552 ***
Honduras
[TR.sub.t]        -2.0091     -3.1213     -3.8804 *       -4.4064 *
[EC.sub.t]        -1.0752     -2.0968     -4.1316 *       -4.7148 *
Hong Kong Sar China
[TR.sub.t]        -1.1785     -1.3189     -2.6850 ***     -3.8314 **
[EC.sub.t]        -2.2905     -2.1313     -4.1514 *       -4.6741 *
Iceland
[TR.sub.t]        -0.0669     -2.9149     -3.9574 *       -3.6995 **
[EC.sub.t]         1.3877     -1.0638     -2.6858 ***     -4.4322 *
Indonesia
[TR.sub.t]         0.2339     -2.9163     -3.0756 **      -3.2696 ***
[EC.sub.t]        -0.8880     -1.1027     -3.0141 **      -5.4069 *
Ireland
[TR.sub.t]        -0.3663     -2.9986     -3.4761 *       -4.3522 **
[EC.sub.t]        -0.7152     -1.7686     -2.8905 ***     -3.9752 **
Italy
[TR.sub.t]        -0.4589     -2.1827     -3.0526 **      -3.6232 **
[EC.sub.t]        -0.6640     -0.6640     -3.7542 *       -3.5772 ***
Japan
[TR.sub.t]        -0.5783     -1.5631     -3.7380 *       -3.7787 **
[EC.sub.t]        -1.5272     -0.7059     -2.9823 ***     -3.4728 **
Kenya
[TR.sub.t]         0.9276     -2.3376     -3.6645 **      -4.5061 *
[EC.sub.t]        -1.8363     -3.0614     -3.3529 **      -3.3313 ***
Kuwait
[TR.sub.t]        -0.9690     -2.0366     -4.6979 *       -5.2502 *
[EC.sub.t]        -2.3481      0.4619     -4.8638 *       -5.8653 *
Luxembourg
[TR.sub.t]        -0.2836     -2.2064     -4.9548 *       -4.8930 *
[EC.sub.t]        -2.3473     -2.3293     -4.0122 *       -5.6876 *
Mexico
[TR.sub.t]         0.2913     -2.4058     -3.8353 *       -3.8029 **
[EC.sub.t]         0.2726     -1.6751     -4.5094 *       -5.8401 *
Hungary
[TR.sub.t]         1.7100     -1.6508     -3.2192 **      -4.3836 **
[EC.sub.t]        -1.5879     -1.6464     -4.2076 *       -4.1344 **
India
[TR.sub.t]         1.8877     -0.6580     -3.0276 **      -3.8732 **
[EC.sub.t]        -0.0584     -2.1698     -3.4824 **      -3.3593 ***
Iran
[TR.sub.t]        -1.8514     -3.1574     -3.9574 *       -3.8381 **
[EC.sub.t]        -1.7349     -2.6435     -4.8904 *       -4.8000 *
Israel
[TR.sub.t]         0.2725     -3.0813     -4.7457 *       -4.6242 *
[EC.sub.t]        -1.3830     -1.3627     -2.6706 ***     -3.9254 **
Jamaica
[TR.sub.t]        -0.9943     -1.0985     -3.0749 **      -3.3349 ***
[EC.sub.t]        -0.5598     -2.9249     -2.9871 ***     -3.9866 **
Jordan
[TR.sub.t]         1.6131     -1.0977     -3.5064 **      -4.1582 **
[EC.sub.t]        -1.6982     -2.4034     -3.9477 *       -3.7925 **
South Korea
[TR.sub.t]        -0.4298     -2.3466     -3.7693 *       -3.7279 **
[EC.sub.t]        -1.1716     -1.7710     -3.3229 **      -3.2994 ***
Morocco
[TR.sub.t]        -0.9696     -2.0819     -4.3410 *       -4.1784 **
[EC.sub.t]        -0.9635     -2.1519     -5.0387 *       -5.2066 *
Nepal
[TR.sub.t]        -2.3691     -1.8741     -3.7489 *       -4.3319 *
[EC.sub.t]         0.4621     -1.3866     -3.7507 *       -4.3404 *
Mozambique
[TR.sub.t]         0.3713     -0.5526     -3.1407 **      -3.3170 ***
[EC.sub.t]        -2.2439     -1.5365     -3.5940 **      -3.7322 **
Netherland The
[TR.sub.t]        -1.4168     -3.2000     -3.8649 *       -3.9471 **
[EC.sub.t]        -2.4361     -2.8255     -5.0101 *       -4.9431 *
Nicaragua
[TR.sub.t]        -0.4710     -1.1263     -3.3732 **      -3.3756 ***
[EC.sub.t]        -1.5720     -1.9819     -4.6927 *       -4.9537 *
Norway
[TR.sub.t]        -1.1537     -2.6473     -4.9267 *       -4.7619 *
[EC.sub.t]        -1.4857     -2.6535     -3.7932 *       -3.6945 **
Pakistan
[TR.sub.t]        -0.8509     -1.5699     -3.6078 **      -3.7826 **
[EC.sub.t]        -0.7991     -1.2641     -3.6304 **      -3.6256 **
Paraguay
[TR.sub.t]        -1.0733     -1.8795     -3.3666 **      -3.2948 ***
[EC.sub.t]        -1.9243     -1.5327     -3.4150 **      -3.5757 ***
Philippines
[TR.sub.t]         0.0850     -2.4948     -2.9139 ***     -4.0941 **
[EC.sub.t]        -1.0685     -0.8958     -2.7434 ***     -5.7293 *
Senegal
[TR.sub.t]         0.3681     -1.9134     -3.9852 *       -4.0835 **
[EC.sub.t]        -2.0357     -1.7417     -3.7402 *       -4.0870 **
Sweden
[TR.sub.t]        -0.2027     -3.2173     -3.6094 **      -3.5278 ***
[EC.sub.t]        -2.3509     -2.2029     -3.7852 *       -4.1207 **
Spain
[TR.sub.t]        -2.6228     -2.9807     -2.9065 ***     -3.9750 **
[EC.sub.t]         0.3351     -2.5762     -3.3364 **      -3.6564 **
Sudan
[TR.sub.t]         0.9521     -0.2051     -2.6364 ***     -3.7561 **
[EC.sub.t]         0.0171     -1.6685     -4.6910 *       -5.0355 *
New Zealand
[TR.sub.t]        -1.0605     -2.9833     -5.2135 *       -5.1376 *
[EC.sub.t]        -1.7181     -0.4779     -3.0886 **      -3.3346 ***
Nigeria 62
[TR.sub.t]        -0.1775     -2.4375     -3.5531 **      -3.9467 **
[EC.sub.t]        -1.7124     -2.4091     -4.8954 *       -4.7717 *
Oman
[TR.sub.t]         0.5709     -1.9620     -4.7076 *       -5.4118 *
[EC.sub.t]        -1.6655     -1.1611     -3.2912 **      -3.8308 **
Panama
[TR.sub.t]        -0.0274     -2.9196     -3.6502 **      -3.7050 **
[EC.sub.t]        -1.4526     -2.1700     -3.5667 **      -3.5796 ***
Peru
[TR.sub.t]         0.9379     -1.2987     -4.1376 *       -4.8637 *
[EC.sub.t]        -2.4168     -1.6216     -3.0831 **      -3.8628 **
Portugal
[TR.sub.t]        -0.9716     -1.9043     -3.1984 **      -3.7547 **
[EC.sub.t]        -1.4205     -0.5693     -3.0971 **      -3.4068 ***
Saudi Arabia
[TR.sub.t]        -1.1196     -3.0603     -2.9303 ***     -3.8555 **
[EC.sub.t]        -0.4166     -2.4292     -4.3369 *       -4.4657 *
South Africa
[TR.sub.t]        -0.1611     -2.2382     -3.3540 **      -3.5337 ***
[EC.sub.t]        -2.4185     -2.7120     -3.9703 *       -3.8643 **
Switzerland
[TR.sub.t]        -0.5370     -2.1945     -3.0437 **      -3.6199 **
[EC.sub.t]        -2.1958     -2.3868     -3.8958 *       -4.1728 **
Thailand
[TR.sub.t]        -0.6347     -1.8510     -2.9256 ***     -3.8709 **
[EC.sub.t]        -0.6523     -2.1115     -2.9460 ***     -3.2717 ***
Syrian Arab Rep
[TR.sub.t]         0.7897     -2.2773     -3.2714 **      -3.7719 **
[EC.sub.t]        -1.3196     -0.1094     -3.9862 *       -4.2562 **
Togo
[TR.sub.t]        -1.6974     -2.0971     -3.2771 **      -3.4455 ***
[EC.sub.t]        -0.6940     -2.2815     -3.7204 **      -3.6245 **
Tunisia
[TR.sub.t]         0.2968     -2.9650     -2.6946 ***     -3.8919 **
[EC.sub.t]        -0.0885     -2.2401     -3.8989 *       -3.6826 **
United Kingdom
[TR.sub.t]         0.2412     -3.2119     -2.7876 ***     -3.2986 ***
[EC.sub.t]        -1.7197     -0.5494     -3.4085 **      -4.1409 **
Uruguay
[TR.sub.t]        -0.1814     -2.6080     -3.0855 **      -3.7887 **
[EC.sub.t]        -2.3534     -3.0691     -4.1359 *       -4.1451 **
Venezuela R.B.De
[TR.sub.t]         0.1327     -2.2907     -3.9118 *       -4.8369 *
[EC.sub.t]        -1.8629     -1.8146     -3.5727 **      -3.4811 ***
Zambia
[TR.sub.t]         0.7516      0.3288     -3.4925 **      -4.2436 **
[EC.sub.t]        -1.5577     -0.5170     -3.8687 *       -4.4820 *
Trinidad and Tobago
[TR.sub.t]         1.0311     -0.9596     -2.8083 **      -4.8930 *
[EC.sub.t]         1.4450     -0.9133     -3.1422 **      -3.4384 ***
Turkey
[TR.sub.t]        -0.4813     -3.1314     -1.9825 *       -4.7570 *
[EC.sub.t]        -1.0464     -2.1727     -3.6186 **      -3.5759 ***
United Arab Emirates
[TR.sub.t]         1.1937     -2.0504     -2.7599 ***     -3.7995 **
[EC.sub.t]        -2.4012     -1.6495     -3.6501 **      -4.0875 **
United States
[TR.sub.t]        -0.5591     -2.7876     -4.2063 *       -3.9376 **
[EC.sub.t]        -2.4541     -1.7094     -5.8708 *       -5.6874 *
Vietnam
[TR.sub.t]        -1.2282     -2.2356     -5.6683 *       -5.7772 *
[EC.sub.t]         1.6287     -0.7176     -3.7120 **      -4.7837 *
Zimbabwe
[TR.sub.t]        -1.6008     -1.6471     -3.1144 **      -3.4239 ***
[EC.sub.t]        -1.1851     -2.0258     -4.1822 *       -4.2352 **

Note: *, ** and *** denote significant at 1 percent, 5 percent and
10 percent levels respectively.

Table 2
Johansen Cointegration Test

Country                       Likelihood    5% critical   P-value
Algeria                          Ratio         Value

R = 0                          348179 *       25.8721     0.0030
R [less than or equal to] 0     5.09129       12.5179     0.5833
Argentina
R = 0                         27.1434 **      25.8721     0.0346
R [less than or equal to] 0     6.42493       12.5179     0.4083
Austria
R = 0                         27.04634 *      25.8721     0.0094
R [less than or equal to] 0    4.400725       12.5179     0.1968
Bangladesh
R = 0                          28.7918 *      25.8721     0.0210
R [less than or equal to] 0     4.95061       12.5179     0.6035
Benin
R = 0                          41.7722 *      25.8721     0.0003
R [less than or equal to] 0    15.0975 *      12.5179     0.0181
Botswana
R = 0                         27.4591 **      25.8721     0.0315
R [less than or equal to] 0    6.463937       12.5179     0.4038
Brunei Darrulsalm
R = 0                         29.4351 **      25.8721     0.0172
R [less than or equal to] 0     9.58154       12.5179     0.1474
Cameroon
R = 0                           24.3665       25.8721     0.0761
R [less than or equal to] 0     9.47495       12.5179     0.1531
Chili
R = 0                          31.5805 *      25.8721     0.0087
R [less than or equal to] 0     8.96315       12.5179     0.1826
Angola
R = 0                           18.4636       25.8721     0.3136
R [less than or equal to] 0     7.45470       12.5179     0.2995
Australia
R = 0                         29.8304 **      25.8721     0.0152
R [less than or equal to] 0     8.00144       12.5179     0.2516
Albania
R = 0                          33.7549 *      25.8721     0.0042
R [less than or equal to] 0     7.23212       12.5179     0.3209
Belgium
R = 0                         26.6517 **      25.8721     0.0400
R [less than or equal to] 0     7.11880       12.5179     0.3323
Bolivia
R = 0                          66.8464 *      25.8721     0.0000
R [less than or equal to] 0     13.1493       12.5179     0.0392
Brazil
R = 0                           13.7969       25.8721     0.6743
R [less than or equal to] 0     3.11117       12.5179     0.8631
Bulgaria
R = 0                           21.5356       25.8721     0.1578
R [less than or equal to] 0     3.88762       12.5179     0.7583
Canada
R = 0                         26.8541 **      25.8721     0.0377
R [less than or equal to] 0     12.1440       12.5179     0.0577
China
R = 0                         25.9354 **      25.8721     0.0491
R [less than or equal to] 0     8.62820       12.5179     0.2045
Colombia
R = 0                         26.9458 **      25.8721     0.0367
R [less than or equal to] 0     7.87041       12.5179     0.2624
Congo Rep
R = 0                           13.0347       25.8721     0.7355
R [less than or equal to] 0     2.38065       12.5179     0.9406
Costa Rica
R = 0                         26.6582 **      25.8721     0.0399
R [less than or equal to] 0     5.27551       12.5179     0.5573
Cuba
R = 0                          35.5558 *      25.8721     0.0023
R [less than or equal to] 0     8.0965        12.5179     0.2439
Denmark
R = 0                          36.5301 *      25.8721     0.0016
R [less than or equal to] 0   13.6372 **      12.5179     0.0324
Ecuador
R = 0                          49.3521 *      25.8721     0.0000
R [less than or equal to] 0   13.7689 **      12.5179     0.0307
El Salvador
R = 0                          35.1654 *      25.8721     0.0026
R [less than or equal to] 0     12.2436       12.5179     0.0555
Finland
R = 0                         26.9650 **      25.8721     0.0365
R [less than or equal to] 0     6.82323       12.5179     0.3633
Gabon
R = 0                          30.0153 *      25.8721     0.0144
R [less than or equal to] 0     11.7234       12.5179     0.0676
Greece
R = 0                         28.2878 **      25.8721     0.0245
R [less than or equal to] 0     8.29920       12.5179     0.2282
Congo Dem Rep
R = 0                           11.5926       25.8721     0.8392
R [less than or equal to] 0     3.06221       12.5179     0.8691
Saudi Arabia
R = 0                          35.8987 *      25.8721     0.0020
R [less than or equal to] 0    17.0467 *      12.5179     0.0082
Cote D Ivories
R = 0                         27.6100 **      25.8721     0.0301
R [less than or equal to] 0     4.79881       12.5179     0.6254
Cyprus
R = 0                         29.5951 **      25.8721     0.0164
R [less than or equal to] 0   12.9237 **      12.5179     0.0427
Dominican Rep
R = 0                          41.7294 *      25.8721     0.0003
R [less than or equal to] 0     9.29973       12.5179     0.1627
Egypt
R = 0                          35.8685 *      25.8721     0.0021
R [less than or equal to] 0     6.10382       12.5179     0.4472
Ethiopia
R = 0                         30.3543 **      25.8721     0.0129
R [less than or equal to] 0     5.16437       12.5179     0.5729
France
R =0                           34.3356 *      25.8721     0.0035
R [less than or equal to] 0     6.76451       12.5179     0.3697
Ghana
R = 0                          35.1224 *      25.8721     0.0027
R [less than or equal to] 0   14.1094 **      12.5179     0.0268
Guatemala
R = 0                         29.5195 **      25.8721     0.0168
R [less than or equal to] 0     10.5420       12.5179     0.1046
Honduras
R = 0                         26.0812 **      25.8721     0.0471
R [less than or equal to] 0     10.9387       12.5179     0.0905
Hungary
R = 0                          44.9969 *      25.8721     0.0001
R [less than or equal to] 0     8.98506       12.5179     0.1813
India
R = 0                         26.1574 **      25.8721     0.0461
R [less than or equal to] 0     4.72569       12.5179     0.6361
Iran
R = 0                          37.4250 *      25.8721     0.0012
R [less than or equal to] 0     9.92483       12.5179     0.1306
Israel
R = 0                           24.6479       25.8721     0.0704
R [less than or equal to] 0     4.03627       12.5179     0.7368
Jamaica
R = 0                         29.4438 **      25.8721     0.0172
R [less than or equal to] 0     7.55742       12.5179     0.2900
Jordan
R = 0                          33.1366 *      25.8721     0.0052
R [less than or equal to] 0     3.17938       12.5179     0.8545
South Korea
R = 0                         27.3817 **      25.8721     0.0322
R [less than or equal to] 0     8.74030       12.5179     0.1970
Luxemburg
R = 0                          40.8911 *      25.8721     0.0003
R [less than or equal to] 0    19.2744 *      12.5179     0.0032
Morocco
R = 0                         29.1988 **      25.8721     0.0186
R [less than or equal to] 0     6.63904       12.5179     0.3837
Hong Kong
R = 0                          37.9506 *      25.8721     0.0010
R [less than or equal to] 0     7.72672       12.5179     0.2748
Iceland
R = 0                          38.8020 *      25.8721     0.0007
R [less than or equal to] 0     5.81125       12.5179     0.4847
Indonesia
R = 0                          31.2241 *      25.8721     0.0098
R [less than or equal to] 0   12.2892 **      12.5179     0.0546
Ireland
R = 0                          34.3030 *      25.8721     0.0035
R [less than or equal to] 0     7.14944       12.5179     0.3292
Italy
R = 0                          17.09164       25.8721     0.4081
R [less than or equal to] 0    4.836427       12.5179     0.6200
Japan
R = 0                          39.5565 *      25.8721     0.0006
R [less than or equal to] 0     10.5050       12.5179     0.1060
Kenya
R = 0                           17.3930       25.8721     0.3862
R [less than or equal to] 0     6.66917       12.5179     0.3803
Kuwait
R = 0                         28.2335 **      25.8721     0.0250
R [less than or equal to] 0     9.24276       12.5179     0.1659
Mexico
R = 0                          48.3444 *      25.8721     0.0000
R [less than or equal to] 0     6.1009        12.5179     0.4476
Mozambique
R = 0                         31.0356 **      25.8721     0.0104
R [less than or equal to] 0     10.8260       12.5179     0.0943
Nepal
R = 0                         27.6112 **      25.8721     0.0301
R [less than or equal to] 0     2.17146       12.5179     0.9572
New Zealand
R = 0                         28.1404 **      25.8721     0.0257
R [less than or equal to] 0     8.54960       12.5179     0.2100
Nigeria
R = 0                          31.4737 *      25.8721     0.0090
R [less than or equal to] 0     8.19985       12.5179     0.2358
Oman
R = 0                         26.4988 **      25.8721     0.0418
R [less than or equal to] 0     8.58027       12.5179     0.2078
Panama
R = 0                           21.1596       25.8721     0.1728
R [less than or equal to] 0     8.20377       12.5179     0.2355
Peru
R = 0                         26.0875 **      25.8721     0.0470
R [less than or equal to] 0     8.41322       12.5179     0.2198
Portugal
R = 0                           12.4912       25.8721     0.7769
R [less than or equal to] 0     3.69726       12.5179     0.7854
Spain
R = 0                          35.3192 *      25.8721     0.0025
R [less than or equal to] 0     10.2042       12.5179     0.1182
Sweden
R = 0                          31.8140 *      25.8721     0.0081
R [less than or equal to] 0     6.4377        12.5179     0.4068
Syrian Arab Rep
R = 0                         29.8728 **      25.8721     0.0150
R [less than or equal to] 0     11.4533       12.5179     0.0748
Netherland The 62
R = 0                         26.4791 **      25.8721     0.0420
R [less than or equal to] 0     11.6056       12.5179     0.0707
Nicaragua
R = 0                           11.8624       25.8721     0.8214
R [less than or equal to] 0     2.8651        12.5179     0.8922
Norway
R = 0                         28.8942 **      25.8721     0.0204
R [less than or equal to] 0     10.5826       12.5179     0.1031
Pakistan
R = 0                           18.0948       25.8721     0.3376
R [less than or equal to] 0     3.5568        12.5179     0.8048
Paraguay
R = 0                          35.5854 *      25.8721     0.0023
R [less than or equal to] 0    14.3679 *      12.5179     0.0242
Philippines
R = 0                           10.9235       25.8721     0.8795
R [less than or equal to] 0     1.93863       12.5179     0.9723
South Africa
R = 0                         31.1438 **      25.8721     0.0100
R [less than or equal to] 0     4.3126        12.5179     0.6965
Sudan
R = 0                           20.9619       25.8721     0.1811
R [less than or equal to] 0     7.2129        12.5179     0.3228
Switzerland
R = 0                         27.5750 **      25.8721     0.0304
R [less than or equal to] 0     7.2930        12.5179     0.3149
Thailand
R = 0                          39.8339 *      25.8721     0.0005
R [less than or equal to] 0     6.4373        12.5179     0.4069
Togo
R = 0                          48.6538 *      25.8721     0.0000
R [less than or equal to] 0     5.0368        12.5179     0.5911
Tunisia
R = 0                          44.0057 *      25.8721     0.0001
R [less than or equal to] 0   16.1203 **      12.5179     0.0120
United Kingdom
R = 0                          44.3407 *      25.8721     0.0001
R [less than or equal to] 0     7.7262        12.5179     0.2748
Uruguay
R = 0                          35.8733 *      25.8721     0.0020
R [less than or equal to] 0     5.38711       12.5179     0.5418
Venezuela R.B.De
R = 0                         30.9671 **      25.8721     0.0106
R [less than or equal to] 0   12.8779 **      12.5179     0.0435
Zambia
R = 0                         30.39876 **     25.8721     0.0127
R [less than or equal to] 0    2.449747       12.5179     0.9345
Senegal
R = 0                         31.1438 **      25.8721     0.0100
R [less than or equal to] 0     4.3126        12.5179     0.6965
Trinidad and Tobago
R = 0                         27.7872 **      25.8721     0.0286
R [less than or equal to] 0     9.6121        12.5179     0.1459
Turkey
R = 0                         30.0648 **      25.8721     0.0141
R [less than or equal to] 0     6.6956        12.5179     0.3773
United Arab Emirates
R = 0                          33.2987 *      25.8721     0.0049
R [less than or equal to] 0     6.3311        12.5179     0.4194
United States
R = 0                          31.4441 *      25.8721     0.0091
R [less than or equal to] 0     1.6455        12.5179     0.9861
Vietnam
R = 0                         26.1699 **      25.8721     0.0459
R [less than or equal to] 0     8.0407        12.5179     0.2484
Zimbabwe
R = 0                           24.9006       25.8721     0.0657
R [less than or equal to] 0     10.0065       12.5179     0.1269

Note: * and ** denote rejection of null hypothesis at 1 percent and
5 percent levels of significance respectively.

Table 3
Panel Unit Root Test

IPS TEST

                      Level                   1st Difference

                         Trend and                    Trend and
Variables    Intercept   Intercept     Intercept      Intercept

[TR.sub.t]    10.5763     -1.1019      -19.8147 *     -16.6784 *
[EC.sub.t]    2.5184      0.6182       -21.5562 *     -17.8725 *

                                     LLC TEST

                      Level                   1st Difference

                         Trend and                    Trend and
Variables    Intercept   Intercept     Intercept      Intercept

[TR.sub.t]    5.6390      -0.4516      -19.1851 *     -16.5538 *
[EC.sub.t]    1.7180      3.4397       -16.4287 *     -13.5677 *

                                   MW(ADF) TEST

                      Level                   1st Difference

                         Trend and                    Trend and
Variables    Intercept   Intercept     Intercept      Intercept

[TR.sub.t]    30.9469    182.3521      366.570 *      296.0253 *
[EC.sub.t]   164.2160    200.3711      563.351 *      445.5541 *

                                    MW(PP) TEST

                      Level                   1st Difference

                         Trend and                    Trend and
Variables    Intercept   Intercept     Intercept      Intercept

[TR.sub.t]    32.2558    178.6561     1064.9488 *      895.8082
[EC.sub.t]   169.0261    196.1862     1471.0689 *     1282.0323 *

Note: * Denotes rejection of null hypothesis at 1 percent
significance level.

Table 4
Panel Cointegration Test

Hypotheses         Likelihood Ratio   1% Critical Value

R = 0                  5.9035 *             2.45
R [less than
 or equal to] 0         0.9523

Note: * Denotes rejection of null hypothesis at 1 percent
significance level.

Table 5
Panel Cointegration Estimates

                   Pooled Mean                     Hausman
Variables          Group (PMG)    Mean Group(MG)   Test (6)

                              High Income Panel (7)

[TR.sub.t]           0.860 *         1.315 **        3.31
                     (0.000)         (0.041)       (0.191)
[TR.sup.2.sub.t]     -0.015 *       -1.688 **
                     (0.000)         (0.054)

                              Middle Income Panel

[TR.sub.t]          -0.023 **       -0.191 ***       1.45
                     (0.014)         (0.063)       (0.484)
[TR.sup.2.sub.t]     0.003 *         0.116 **
                     (0.000)         (0.043)

                                Low Income Panel

[TR.sub.t]           -1.493 *       -2.827 **        1.68
                     (0.000)         (0.023)       (0.321)
[TR.sup.2.sub.t]     0.0387 *        0.114 **
                     (0.000)         (0.030)

Note: *, ** and *** show significance at 1 percent, 5 percent
and 10 percent levels respectively.

Table 6
Non-Homogenous and Homogenous Causality

                         Non-homogenous Causality
Dependent
Variables         In [TR.sub.t]        In [EC.sub.t]

In [TR.sub.t]           --           Causality Exists *
In [EC.sub.t]   Causality Exists *           --

                        Homogenous Causality
Dependent
Variables           [TR.sub.t]        [EC.sub.t]

In [TR.sub.t]           --           No Causality
In [EC.sub.t]   Causality Exists *        --

Note: * Represents significance at 1 percent level.

Table 7
Homogenous and Non-homogenous Causality

                         Homogenous Causality

Variables               High Income Countries

               In [TR.sub.t]        In [EC.sub.t]

[TR.sub.t]           --           Causality Exists *
[EC.sub.t]   Causality Exists *           --

Variables             Middle Income Countries

               In [TR.sub.t]        In [EC.sub.t]

[TR.sub.t]           --           Causality Exists *
[EC.sub.t]   Causality Exists *           --

Variables              Low Income Countries

               In [TR.sub.t]        In [EC.sub.t]

[TR.sub.t]           --           Causality Exists *
[EC.sub.t]   Causality Exists *

                      Non-homogenous Causality

Variables              High Income Countries

                 [TR.sub.t]           [EC.sub.t]

[TR.sub.t]           --              No Causality
[EC.sub.t]   Causality Exists *

Variables            Middle Income Countries

                 [TR.sub.t]           [EC.sub.t]

[TR.sub.t]                        Causality Exists *
[EC.sub.t]   Causality Exists *

Variables            Low Income Countries

                 [TR.sub.t]           [EC.sub.t]

[TR.sub.t]                        Causality Exists *
[EC.sub.t]   Causality Exists *

Note: * Represents the significance at 1 percent level.

Table 8
Heterogeneous Causality

Country         Variables        [TR.sub.t]             [EC.sub.t]

Algeria        [TR.sub.t]           ----               No Causality
               [EC.sub.t]    Causality exists *             --
Angola         [TR.sub.t]            --                No Causality
               [EC.sub.t]    Causality exists *            ----
Argentina      [TR.sub.t]            --                No Causality
               [EC.sub.t]    Causality exists *             --
Australia      [TR.sub.t]            --                No Causality
               [EC.sub.t]    Causality exists *             --
Austria        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Albania        [TR.sub.t]            --             Causality exists *
               [EC.sub.t]   Causality exists ***            --
Bangladesh     [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]       No Causality                --
Belgium        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Benin          [TR.sub.t]            --             Causality exist **
               [EC.sub.t]       No Causality                --
Bolivia        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Botswana       [TR.sub.t]            --                No Causality
               [EC.sub.t]    Causality exists *             --
Brazil         [TR.sub.t]            --                No Causality
               [EC.sub.t]    Causality exists *             --
Brunei         [TR.sub.t]            --                No Causality
  Darussalam   [EC.sub.t]       No Causality                --
Bulgaria       [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Cameroon       [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Canada         [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Chile          [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
China          [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Colombia       [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Congo Dem Rep  [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Congo Rep      [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Costa Rica     [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Cote D'Ivoire  [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]     Causality exist *             --
Cuba           [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Cyprus         [TR.sub.t]            --             Causality exist **
               [EC.sub.t]     Causality exist *             --
Denmark        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Dominican Rep  [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Ecuador        [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Egypt          [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]     Causality exist *             --
El Salvador    [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Ethiopia       [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Finland        [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
France         [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Gabon          [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]     Causality exist *             --
Ghana          [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Greece         [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Guatemala      [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Honduras       [TR.sub.t]            --             Causality exist **
               [EC.sub.t]     Causality exist *             --
Hong Kong      [TR.sub.t]            --             Causality exist *
               [EC.sub.t]    Causality exist ***            --
Hungary        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Iceland        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
India          [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Indonesia      [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Iran           [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Ireland        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Israel         [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Italy          [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Jamaica        [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
Japan          [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Jordan         [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Kenya          [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
South Korea    [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Kuwait         [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
Luxemburg      [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Mexico         [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Morocco        [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
Mozambique     [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Nepal          [TR.sub.t]            --                No Causality
               [EC.sub.t]    Causality exist **             --
The            [TR.sub.t]            --             Causality exist *
  Netherlands  [EC.sub.t]       No Causality                --
New Zealand    [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Nicaragua      [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Nigeria        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Norway         [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
Oman           [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Pakistan       [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Panama         [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
Paraguay       [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]       No Causality                --
Peru           [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]       No Causality                --
Philippines    [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]       No Causality                --
Portugal       [TR.sub.t]            --                No Causality
               [EC.sub.t]    Causality exist **             --
Saudi Arabia   [TR.sub.t]            --             Causality exist **
               [EC.sub.t]     Causality exist *             --
Senegal        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
South Africa   [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Spain          [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Sudan          [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Sweden         [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]       No Causality                --
Switzerland    [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Syria          [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Thailand       [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
Togo           [TR.sub.t]            --            Causality exist ***
               [EC.sub.t]    Causality exist ***            --
Trinidad       [TR.sub.t]            --             Causality exist *
  and Tobago   [EC.sub.t]       No Causality                --
Tunisia        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Turkey         [TR.sub.t]            --                No Causality
               [EC.sub.t]     Causality exist *             --
United         [TR.sub.t]            --                No Causality
  Kingdom      [EC.sub.t]       No Causality                --
United Arab    [TR.sub.t]            --             Causality exist *
  Emirates     [EC.sub.t]       No Causality                --
Uruguay        [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
Unites States  [TR.sub.t]            --             Causality exist *
               [EC.sub.t]     Causality exist *             --
Venezuela      [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Vietnam        [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --
Zambia         [TR.sub.t]            --             Causality exist *
               [EC.sub.t]       No Causality                --
Zimbabwe       [TR.sub.t]            --                No Causality
               [EC.sub.t]       No Causality                --

Note: *, ** and *** represent significance at 1 percent, 5 percent and
10 percent levels respectively.