3.1 Data
This research explore the impact of CO2 emission, inflow of foreign direct investment, renewable energy usage, financial development and economic growth in the world, world high income countries, upper middle income countries and lower middle income countries for the time 1980 to 2018. The study used panel data where all these variables data have been gathered from the World Bank, world development indicator. The variables used in this study are FDI (foreign direct investment) which has been taken as a percent of Gross domestic product GDP, international tourism, carbon emission which is calculated as metric tons per capita, renewable energy consumption (calculated as a percent of total ending energy), financial development by bank proxy by domestic credit to private sector by bank calculated as a percent of GDP where the dependent variable is per capita GDP. Moreover, controls variables have been used in the study are urban population, total government expenditure, trade openness and merchandise trade (% of GDP).
3.2 Econometric Models and Methods
The current study use different models which includes fixed effect, OLS, GMM and System GMM models to examine the effect of FDI, international tourism receipts, renewable energy consumption, CO2 emission and financial development FD by bank on economic development in income group’s economics. The system GMM model most effective estimator therefore used, because OLS & fixed-effects models are not efficient and may cause several econometrics problems. We used these static models to compare this research with previous research that found different results compare the results. In addition, we also use a differential GMM model, where the differential GMM uses the first-order difference of the regressor and converts the regression of the dependent variable, which can handle country specific effects and make the regressor time constant. In this model, the first differences of lagged dependent variable are using to eliminate the problems of autocorrelation. (Arellano & Bover, 1995) It shows that this model may also give inefficient conclusions due to the deliberate nature of the independent variables. Therefore, the system GMM model has considered by different scholars and pointed out that the two step GMM estimator is the most effective estimator (Law & Azman-Saini, 2012). The current research finally focus on system GMM model to examine the impact of the study variables on economic growth in income groups.
The empirical model can be explained as follows;
Where EG is the dependent variables used to represent economic growth, foreign direct investment FDI is the inflow of FDI calculated as net inflows as percent of GDP Gross domestic product , ITR is international tourism, RE represents the consumption of renewable energy, CO2 represents carbon dioxide, FD represents the financial development of the bank, and EGit-1 is the first lag of the left-hand dependent variable, used as an explanation in the equation to quantify the effect of the previous year The variable in the last year, Xit represents the direction of the control variable, and the hypothesis will affect our left variable. The control variables of the study are urban population, government expenditure, trade openness and commodity trade. The subscripts in the equation specify (i = 1. . . N) and (t = 1980 . . . 2018) index country & time respectively.
3.4 Findings and Discussions
The following section illustrates that results on the effect of CO2 emission, FD financial development, tourism, RE and foreign direct investment FDI on economic growth for the whole sample. In the table below, column 1 illustrates the list of independent and control variables of the study, column 2 present of OLS model the results, column 3, 4 and column 5 present the results of fixed effect, system GMM and GMM models respectively. The result of GMM indicates that the lagged dependent variables are highly statistically significant which shows the suitability of the dynamic models both system and difference GMM. Similarly, the Sargan test and AR -1 & AR-2 p values also indicates the suitability of the models.
Table 1: Aggregate analysis for the whole sample
Economic Growth
|
Model-1
|
Model-2
|
Model-3
|
Model-4
|
Foreign direct invest
|
0.083***
|
0.081**
|
0.061***
|
0.068***
|
|
(0.023)
|
(0.031)
|
(0.001)
|
(0.000)
|
Tourism
|
-0.070***
|
-0.076
|
-0.019*
|
-0.089***
|
|
(0.023)
|
(0.060)
|
(0.010)
|
(0.000)
|
Carbon emission
|
0.011
|
0.492***
|
-0.026***
|
0.028***
|
|
(0.024)
|
(0.173)
|
(0.008)
|
(0.001)
|
Renewable energy
|
-0.004***
|
-0.004
|
-0.011***
|
-0.003***
|
|
(0.001)
|
(0.007)
|
(0.001)
|
(0.000)
|
Financial development
|
-0.177***
|
-0.344***
|
-0.199***
|
-0.188***
|
|
(0.040)
|
(0.081)
|
(0.012)
|
(0.001)
|
Urban Population
|
2.500***
|
-3.120
|
-5.890***
|
1.620***
|
|
(7.321)
|
(5.130)
|
(3.181)
|
(0.000)
|
Trade Openness
|
0.089
|
0.233
|
0.786***
|
0.187***
|
|
(0.107)
|
(0.340)
|
(0.019)
|
(0.006)
|
Govt. Expenditure
|
-0.065***
|
-0.023
|
-0.007**
|
-0.063***
|
|
(0.0187)
|
(0.037)
|
(0.003)
|
(0.000)
|
Merchandize trade
|
0.252**
|
0.423
|
0.342***
|
0.057***
|
|
(0.103)
|
(0.297)
|
(0.020)
|
(0.003)
|
L.GDP per capita
|
|
|
-0.080***
|
0.251***
|
|
|
|
(0.001)
|
(0.000)
|
Constant
|
1.824***
|
-3.509**
|
|
2.262***
|
|
(0.458)
|
(1.774)
|
|
(0.027)
|
|
|
|
|
|
Observations
|
1,215
|
1,215
|
646
|
981
|
R-squared
|
0.118
|
0.051
|
|
|
Number of id
AR1
AR2
Sargan test
|
|
153
|
128
-3.64(0.000)
-2.24(0.025)
725.55(0.000)
|
151
-4.22 (0.000)
-0.31(0.759)
888.71(0.000)
|
Source: Own calculation
Note: OLS is the ordinary least squares method, and FE, SGMM & GMM are fixed effects, which are the generalized method of moments and the generalized method of moments of the system, respectively. The standard error is shown in parentheses, and the significance level is shown by *, **, *** at 1, 5 and 10 percent respectively. AR1 and AR2 are the Arellano and Bond tests.
The table above indicates that the estimated co efficient of FDI is significant highly at one percent significance level and its relationship with economic growth is positive which illustrates that increase in FDI enhance economic growth in the global panel. For instance, the System GMM model results indicate that a unit rise in the level of FDI foreign direct investment will augment economic growth by 0.068 percent in the globe. It’s well known in the preceding literature that FDI is one of the main drivers of economic growth in most of countries and our findings also suggest that FDI is the essential driver of economic growth in the global panel of this study. The current study results are further reinforced by the study results of (Choong & Lim, 2009); (Hong & Sun, 2011); and (Sokang, 2019) who also statutes that FDI increase economic growth.
Likewise, our study also used tourism and its role in enhancing economic growth where the estimated co efficient of tourism is high significant while its association with economic growth is negative in the global panel which shows that tourism decrease economic growth. Specially the estimated value of system GMM indicates that a percent increase in tourism lowers the level of economic growth by 0.089 % in the panel. The negative results maybe the reason of poor level of tourism in some countries. The present study results are contradictory with the findings of (Fayissa, Nsiah, & Tadasse, 2007).
Similarly, the renewable energy estimated coefficient has found significantly mostly in all models while it is impact on economic growth is negative, which shows that a rise in the use of RE lowers economic growth in the globe while it’s been also argued that renewable energy consumption is beneficial for environmental quality. For instance, the system GMM results shows that a percent increase in the use of renewable energy will reduce economic growth by 0.003% in the globe. Our results further suggest that renewable energy related financing should be promoted to avoid its negative influence on economic growth as economic growth is very important of any country as well the renewable energy consumption is also beneficial to promote environmental quality. They study of (Bilan et al., 2019) also reinforced the findings of the current study where they use DOLS model illustrates that increase in renewable energy RE consumption decrease economic growth.
Likewise, the estimated coefficient of carbon dioxide emission in fixed effect and system GMM models are highly statistically significant and positive which shows that a percent increase in carbon emission will increase economic growth by 0.40 and 0.28 percent. The results illustrates that emission of carbon dioxide enhances the level of economic growth in the globe. The current findings of the study regarding carbon emission is in line with the findings of (Fan & Hossain, 2018), (S. Khan, Peng, & Li, 2019) and (Mehtab et al., 2019). (Muhammad, 2019) also founds that there is positive impact of carbon emission CO2 on economic growth in MENA countries.
Likewise, the estimated co efficient of financial development (FD) by bank which is proxied by credit to private sector is also highly significant at a 1% significance level while in the panel the impact on economic growth is negative. For instance, the results of system GMM indicates that if there is one percent rise in credit to private sector will decrease growth by 0.88 percent in the globe. The FD impact is negative on economic growth maybe the reason of low income countries in the panel with poor financial system. (S. Khan et al., 2019) and (Al-Mulali & Sab, 2012) also statues in their study that financial development decrease economic growth.
Moreover, the coefficient of urban population which is a control variable used in the model is highly statistically significant and positive in OLS and system GMM while the relationship with economic growth is negative only in DGMM. The results of system GMM indicates that growth urban population increase economic growth. Likewise, the coefficient value of trade openness is significant and positive in GMM and system GMM which indicates that trade openness augment economic growth where these results are similar to the previous study results of (Fan, Hossain, Islam, & Yahia, 2019). Similarly, government expenditure is also highly statistically significant in all models except FE while the relationship with negative economic growth which illustrates that government expenditure decrease economic growth in the panel. Likewise, the coefficient of merchandize is significant at 1% level and positive which statutes that merchandize trade enhances the level of economic growth in the panel.
Table 2: Two-step system GMM results for income groups
Economic Growth
|
High-Income Countries
|
Upper-Middle Income
|
Lower-Middle Income
|
Foreign direct invest
|
0.108***
|
-0.050***
|
-0.055***
|
|
(0.003)
|
(0.000)
|
(3.600)
|
Tourism
|
-0.245***
|
0.265***
|
0.0171***
|
|
(0.009)
|
(0.000)
|
(6.730)
|
Carbon emission
|
0.156**
|
0.000***
|
-2.820***
|
|
(0.066)
|
(0.000)
|
(3.860)
|
Renewable energy
|
0.004***
|
-0.143***
|
0.000***
|
|
(0.001)
|
(0.000)
|
(0.000)
|
Financial development
|
-0.003***
|
-0.028***
|
0.072***
|
|
(0.000)
|
(3.340)
|
(3.990)
|
Urban Population
|
1.050
|
7.110***
|
5.150***
|
|
(1.101)
|
(0.000)
|
(0.000)
|
Trade Openness
|
0.002***
|
0.000***
|
0.000***
|
|
(0.000)
|
(0.000)
|
(0.000)
|
Govt. Expenditure
|
-0.071***
|
-0.0248***
|
-0.016***
|
|
(0.005)
|
(0.000756)
|
(0.000)
|
Merchandize trade
|
-0.128**
|
0.000***
|
0.000***
|
L.GDP per capita
|
(0.056)
0.238***
(0.001)
|
(0.000)
-0.110***
(9.600)
|
(0.000)
0.000***
(0.000)
|
Constant
|
5.055***
|
0.000***
|
0.000***
|
|
(0.725)
|
(0.000)
|
(0.000)
|
|
|
|
|
Observations
|
357
|
331
|
280
|
R-squared
|
|
|
|
Number of id
|
50
|
43
|
37
|
|
Source: Own calculation
Note: system generalized method of moments SGMM. And the standard error is shown in parentheses, and the significance level is shown by *, **, *** at 1, 5 and 10% respectively
The two step system result is given in the table above for income groups on the impact of FDI, international tourism receipts, financial development by bank, renewable energy and carbon emission on economic growth where column 1 presents the study variables and column 2, 3 and 4 shows the two step system GMM results of High income countries, upper middle income & lower middle income economies respectively.
The results suggest that the lagged dependent variable, which is economic growth, is significant highly which illustrates that the employed estimator is suitable. The two step-system GMM results indicates that the estimated coefficient of FDI is significant for all income groups where the relationship with economic growth is positive in high income countries which indicates that the inflow of FDI in high income countries enhance economic growth. The positive impact can be the reason of high level of FDI inflow in these countries as high income countries have facilitated the flow of foreign direct investment. However the relationship is negative for upper middle and lower middle income countries which indicate that increase in foreign direct investment decrease economic growth which can be reason of low inflow of FDI. (Mustafa & Santhirasegaram, 2013); (Ozturk, 2007), and (Almfraji & Almsafir, 2014) have found the similar results to the current findings where they shows the positive effect of FDI on economic development.
In addition, the assessed coefficient of tourism for all three income groups bunches economies is significantly significant and the relationship with economic growth is positive for upper, and lower middle of income countries which statues that if there is a unit increase in tourism receipt will increase economic growth by 0.265 and 0.017 percent in upper middle and lower middle income countries respectively. The present study results are reinforced by the study’s findings of (Paramati, Alam, & Chen, 2017) who have also found that tourism enhance economic growth in developed and developing countries. However its negative in high income countries. While it will reduce economic growth by 0.245 percent in high income countries. The present findings are reinforced by the studies of (Paramati et al., 2017) and (L.-j. WANG, YUAN, & HE, 2010) where they also found the positive association.
likewise, the estimated coefficient of tourism for all three income groups economies is extremely statistically significant and also the relationship with economic growth is positive for upper middle and lower middle countries which statues that if there's a percent rise in tourism receipt will increase economic growth by 0.265 and 0.017 percent in upper middle & lower middle income countries respectively. The present findings are reinforced by the study of (Paramati, Alam, & Chen, 2017) who have also found that tourism enhance economic growth in developed and developing countries. However its negative in high income countries. While it'll reduce economic growth by 0.245 percent in high income countries. This outcomes are in line with the preceding studies results of (Paramati et al., 2017) and (L.-j. WANG, YUAN, & HE, 2010) where they also found the positive association.
Similarly, the Carbon emission estimated coefficient is extremely significant and the relationship with economic growth is positive for top income & upper middle income countries which indicates that increase in greenhouse gas emission increase economic growth in these countries. Same results regarding carbon emission on economic process have found by (Fan & Hossain, 2018); (Bilan et al., 2019) and (Issaoui, Toumi, & Touili, 2015). However the carbon emission is negative relationship with economic growth for lower middle income economies which illustrates that increase in carbon emission CO2 in these countries decrease economic growth.
Moreover, renewable energy RE consumption is also found to be highly statistically significant for all income based countries and also the relationship with economic process is positive for top and lower middle income groups which indicates that a percent rise in renewable energy consumption rise economic growth by 0.004 percent. The results indicates that prime income and lower middle income countries have achieved the increased level of renewable energy in total energy budget which shows that these objectives are associated to their renewable energy shares within the final energy consumption. As it's far known that the renewable energy zone need lots of financing and the acquired outcomes shows that excessive profits and lower income nations have enough assets and have supported the development of renewable power intake. By the use of FMOL model, (Bilan et al., 2019) also founds the equal results to our findings. However the renewable energy consumption RE relationship with economic growth is negative for upper middle income countries which indicate that renewable energy consumption decrease economic growth in upper middle income countries.
The result may be due to the substitutability between renewable energy consumption and fossil energy, which reduces the former's energy use when increasing the quantity or when the renewable energy consumption is consumed. It also can be the low level of financing to renewable energy consumption in upper middle income countries. Our results are more reinforced by the study of (Singh, Nyuur, & Richmond, 2019); (Ntanos et al., 2018) who also shows that positive effect of renewable energy consumption on economic development.
Similarly, the model also uses bank private sector credit as an agent for financial development, where the estimated coefficient is found to be significantly positive for economic growth in low- and middle-income countries, and significantly positive for high- and middle-income and high-income countries. More precisely, the value of two step S-GMM indicates for lower middle income countries that if there is extension in private sector credit will boost the growth level by 0.072 percent in these countries. The results indicate that credit extension to private sectors is efficient so that’s why the financial development proxied by private sector credit increases economic growth. While the other income grouped countries results indicates that financial development decrease economic growth which can be the reason of low credit to private sector extension. The current results are in line with findings of (Jalil & Ma, 2008) and (Lenka, 2015). Urban population has been used as a control variable in the model where it’s also significant which indicates that its rise economic growth in upper and lower middle income countries while the value are insignificant for high income countries while have no effect on growth in high income countries. The co -efficient of trade openness is also highly statistically significant and positive for all income grouped countries which indicates that trade openness increase economic growth in all economies. Its further suggests that the trade openness level has been increased and facilitated in these economies. (Butkiewicz & Yanikkaya, 2003) also statues that the level of trade openness rise economic growth. Similarly, government expenditure is also significant and negative for all groups which illustrates that government expenditure in the sample countries negatively affect economic growth. Merchandize trade in the model is also significant highly while its impact on growth in negative high income while positive in upper middle & lower middle income countries which shows that increase in merchandize trade rise economic growth for upper middle -lower middle income economies while decrease economic growth in high income countries.
Table 3: Fixed effect model results on income groups
Economic Growth
|
High-Income Countries
|
Upper-Middle Income
|
Lower-Middle Income
|
Foreign direct invest
|
0.0956*
|
0.021**
|
0.042***
|
|
(0.050)
|
(0.010)
|
(0.012)
|
Tourism
|
-0.132
|
-0.000***
|
-0.000***
|
|
(0.175)
|
(0.000)
|
(5.041)
|
Carbon emission
|
0.263
|
0.389
|
0.267
|
|
(0.446)
|
(0.327)
|
(0.217)
|
Renewable energy
|
-0.011
|
0.007
|
0.006
|
|
(0.017)
|
(0.016)
|
(0.011)
|
Financial development
|
-0.005***
|
-0.009**
|
-0.012**
|
|
(0.002)
|
(0.004)
|
(0.005)
|
Urban Population
|
-9.560
|
7.001
|
-1.480
|
|
(3.680)
|
(1.240)
|
(6.920)
|
Trade Openness
|
-0.000
|
-1.449**
|
1.338**
|
|
(0.005)
|
(0.608)
|
(0.624)
|
Govt. Expenditure
|
-0.109
|
-0.007
|
-0.029
|
|
(0.101)
|
(0.060)
|
(0.051)
|
Merchandize trade
|
0.426
|
1.900***
|
-0.277
|
Constant
|
(0.452)
0.393
(4.503)
|
(0.537)
-4.102
(3.637)
|
(0.579)
-5.901**
(2.648)
|
Observations
|
426
|
400
|
329
|
R-squared
|
0.066
|
44
|
38
|
Number of id
|
50
|
0.071
|
0.101
|
|
Source: Own calculation
Note: is system Generalized technique of moments SGMM. Standard error presented in parenthesis, significance level are presented by *, **, *** at 1, 5 and 10 percent respectively.
Table 3 illustrates the estimated results of fixed effect model on the effect of FDI (foreign direct investment), international tourism receipts, consumption of renewable energy, (CO2) Carbon dioxide and (FD) financial development on economic growth in income based grouped economics where column (1) presents the study variables & column (2),(3) and(4) shows the fixed effect model results of High income countries, upper middle income and lower middle income economies respectively.
The results of fixed effect model in the above given table indicates that the co efficient of FDI is significant at 10% , 5 percent and 1 percent and positive for high income, upper middle income and lower middle income respectively. These results indicate that foreign direct investment increase economic growth positively for all income countries and a rise in foreign direct investment will upsurge economic growth of these economies. (Hoàn et al., 2019), (Makki & Somwaru, 2004) also finds the same results to our study findings.
Moreover, the fixed effect results regarding tourism is significant only for upper middle and lower middle income countries while negative relationship with economic growth which indicates that tourism influence economic growth negatively but significantly. The results is opposing to the results of (Paramati et al., 2017) who have found positive impact of tourism on economic growth. On the other hand the effect of renewable energy RE consumption in the fixed effect model for all income group economies is insignificant which indicates that there is no effect of renewable energy consumption on economic growth in these countries. (Taizeng, Can, Paramati, Fang, & Wu, 2019) and (Fayissa et al., 2007) also found the similar results to our study.
The estimated co efficient of FD financial development in the model is highly statistically significant for all grouped countries while the negative relationship with economic growth. More specifically, if there is a percent rise in credit to private sector will decrease economic growth by 0.005, 0.009 and 0.012 in high income, upper middle income and lower middle income countries respectively. The results further indicate that there is low level of private sector credit extension which contributes negatively to economic growth. The findings are reinforced by the study of (Zahir & Masih, 2018) who founds that financial development positively affect economic growth. Likewise, government expenditure and urban population impact on economic growth is highly insignificant while trade openness is negative significant for upper middle income and positive significant for lower middle income countries. Similarly, merchandize trade is only positively significant for upper middle income countries while insignificant for other groups of countries.