*3.1. Results of Panel ARDL*

Table 2 represents the statistical summary of the dependent and independent variables of Equation (2). It shows that the values of minimum, maximum, kurtosis, standard deviation, average, and skewness of the indicator, as mentioned earlier, have reflected an improved understanding of the data and their distribution within the structure—the outcomes of the correlation matrix present in Table 3.


**Table 2.** Statistical summary.

Source: authors own calculations.

**Table 3.** Correlation matrix.


Source: authors own calculations.

The correlation matrix results find that renewable energy usage (ENC) correlates with the ecological disorder (END). At the same time, all other concerned variables have a weak correlation with the ecological disorder. GDP-square variable (GW2) shows the desired negative association with the ecological disorder. As the application of Panel ARDL apple concerning panel unit root, Table 4 represents the panel unit root results. According to the output, variables are stationary at different levels.


**Table 4.** Panel unit root.

Source: authors own calculations. Note: Parentheses have Probability values.

To view the co-integration association between dependent and a set of independent variables of Equation (2), the study performs the bound test, depicted in Table 5, according to the bounds test results. F-statistics (estimated) is higher than the upper and lower critical value bunds. Thus, bounds test results accepted the co-integration of energy usage and environmental depletion alongside other demographic and economic indicators.

#### **Table 5.** Results of the bound test.


Source: authors own calculations. Note: table CI cited the unrestricted intercept, and no critical trend values of Lower bound at 5% = 2.45 and unrestricted intercept and no trend critical values of Upper bound at 5% = 3.61.

Then, Panel ARDL applied to test the long-run influence of economic upswing on environmental depletion. The Panel ARDL projection for the long run prearranges in Table 6.


**Table 6.** Panel autoregressive-distributed lag (ARDL) (Long Run).

Source: authors own calculations, Note: \*\* 5% and \*\*\* 1% show the statistical significance level.

The Panel ARDL findings confirmed EKC (inverted U-shape curve) for these selected developing Asian economies. Economic upswing (GW) also has a positive association in terms of ecological disorder and found that 0.78 units of CO2 emissions are being generated by the economic upswing to pollute the environment. The results depict the positive influence of GDP growth on carbon dioxide emission in developing economies, evidenced by past research [50]. However, the confirmation of EKC proved by the coefficient value <sup>−</sup>0.31 of GDP-square (GW2). The negative coefficient value of GW<sup>2</sup> represents the negative bond between GDP-square and carbon emission, which confirmed the reduction of carbon dioxide emission in the developing Asian economies. Thus, reducing carbon dioxide emissions due to improved economic upswing has shown the presence of inverted U-shaped EKC in developing economies and confirms the findings of [51,52].

It further finds that renewable energy usage participates negatively in carbon dioxide emissions in these developing economies. The renewable energy usage (ENC) coefficient is −0.26, which indicates that one percent of energy usage is a source of 0.26 carbon emission reduction emission. The coefficient of renewable energy usage is also significant at one percent. Hence, it proved that renewable energy usage growth helps reduce the pollution of these developing economies. The previous studies also establish the same affirmative influence of renewable energy usage on developing economies' carbon dioxide emissions [53,54].

Capital formation (FCF) has shown a role in terms of increasing CO2 emissions. The results show that enhancement in capital formation has increased the environmental depletion in the developing economies. Results suggest that with ceteris paribus, a one percent increase in capital formation is a source of 0.11 percent of carbon dioxide emission, and [51,55] have shown the same evidence in their past study in which the improvement in capital formation was promoting the ecological disorder. The long-run panel ARDL does not significantly affect population growth (PG) on the ecological disorder. According to past research, population growth can be an essential indicator of any country's economy. However, it has no substantial evidence to affect the environmental conditions in the ecological disorder.

According to Table 7, the short-run results are insignificance for the economic upswing and environmental depletion in these selected developing economies. The short-run regressor consists of lag terms of the previous year. Almost all variables were found to be insignificant concerning the depletion of the environment in the preceding year. Thus, the short-run results confirmed that all economic (economic upswing, GDP-square, energy usage, fixed capital formation) and demographic indicators do not affect the carbon dioxide emission of these developing economies of Asia. This short-run analysis estimates through ECM. The coefficient value of ECMit-1 is 0.26, which shows convergence toward the long run from the short run to attain equilibrium condition. The significant negative value has proven the belongings of ECM to form the steadiness by dropping error. It estimates that the unsteadiness or errors are diminished by about 26 percent each year towards the long run from the short run, which helps attain equilibrium conditions among economic upswing, fixed capital formation, energy usage, and environmental depletion in these selected developing Asian economies. Regarding our other socio-economic variables' unemployment rate, we found it positive (as expected), but not statistically significant. Our findings are consistent with the results drawn by [51,56].


**Table 7.** Short-run panel ARDL with Error Correction Model.

Source: authors own calculations, Note: \* 10%, \*\* 5% and \*\*\* 1% show the statistical significance level.

Table 8 demonstrates the results of the augmented Dickey-Fuller (ADF) test. Generally, the average time level has no stationary series, even though all series are stationary having the first difference. If the ADF test did by using the first difference, generally, the insignificant supposition is rejected at the significance level of 1% or 5%. Consequently, this stated that the data are converted into the shape of stationary with the first difference.

**Table 8.** Unit root results of the augmented Dickey-Fuller test (ADF).


Source: Author's own calculation by using E-Views 5. ENG stands for the ecological disorder, ENC shows the energy consumption, EGW is economic growth, EGW<sup>2</sup> is the square of economic growth, FCF is capital formation, PG shows population growth.

The results of the error correction model presented in Table 9.


**Table 9.** Error correction model results.

Source: Author's own calculation by using E-Views 5. Note: \*\*\* denotes 1% significance level. ENG stands for the ecological disorder, ENC shows the energy consumption, EGW is economic growth, EGW2 is the square of economic growth, FCF is capital formation, PG shows population growth.

The ECt−<sup>1</sup> coefficient demonstrates the short-run adjustment rate to the long-run rate, whereas the adjustment rate was noted as 12%; this implies that the 12% imbalance is corrected per year. The numerical coefficient of significance ensures the long-run causality among independent variables and dependent variables.

This study develops three indexes (environmental index (EVI), energy efficiency index (EE), and energy intensity index (EIN). The EVI index develops by using all indicators of Table 1. In contrast, the EE index utilizes energy efficiency and energy consumption as input indicators, and the EIN index has been developed by dividing energy consumption by GDP.
