The study employs annual data over the period 1990–2022 from the World Bank. First, the study graphically plots the data trend to determine whether the form of the relationship between urbanization rate and CO2 emissions per capita is linear. The vertical axis represents CO2 emissions per capita (dependent variable), while the horizontal axis represents urbanization rates (independent variable).

Figure 1 clearly illustrates that the relationship is nonlinear and close to an inverted U-shape. The coordinates in the above figure are (0.73, 2.92), (0.78, 3.69), and (0.91, 1.9) for the beginning, peak, and trough points, respectively. By linking these coordinates, the graph shows an inverted U-shape. This study uses the threshold regression method to estimate the model for three reasons. First, threshold estimation enables us to avoid the collinearity/multicollinearity problem that appears with the quadratic equation (Narayan and Narayan 2010). Second, Fig. 1 shows that the dependent variable (CO2 emissions per capita) does not tend to decline until the urbanization rate reaches the threshold value, which is approximately 0.78. Third, threshold regression is more consistent when different patterns of relationships exist (different regimes and policies). The study follows steps to estimate the relationship between the urbanization rate and share of renewables in electricity production (wind and solar) variables and the dependent variable (CO2 emissions per capita). The first step is conducting the sequential F-statistics-determined threshold test.

Table 1 shows the results of the sequential F-statistics determined threshold. The results show two threshold values for the urbanization rate (threshold repressor), which are (0.78, and 0.91). The threshold variable in this study (urbanization rate) is used to split the sample into three regimes or groups. The first and second groups respectively represent data points before and after the threshold value of the urbanization rate, which is 0.78. The third group represents data points after the trough value of the threshold variable, which is approximately 0.91. The sequential F-statistics test supports the tendency of this study to use the threshold approach for model estimation.

Table 1

Sequential F-statistics determined threshold

Threshold Test | F-statistics | Scaled F-statistics |

0 vs. 1* | 20.1540 | 60.4621 |

1 vs. 2* | 6.4026 | 19.2080 |

2 vs. 3 | 1.5396 | 4.61 |

Threshold Values | **Sequential** | **Repartition** |

1 | 0.783199 | 0.783199 |

2 | 0.911999 | 0.911999 |

Source: Using Eviews 12 |

*Significant at the 0.05 level |

The next step to ensure that the threshold regression will reveal consistent results is conducting the unit root. Table 2 shows the result of the unit root test:

Table 2

Unit root test for the share of renewable in electricity production variable (at level)

| t-Statistics | Prob |

Augmented Dickey-Fuller Test Statistics | -4.5878 | 0.0011 |

Test Critical Value: | 10% level | -3.6891 | |

1% level | -2.9718 | |

5% level | -2.6251 | |

Source: Using Eviews 12 |

According to the results in Table 2, the unit root test of the share of renewables in the electricity production variable (REP) shows that this variable is stationary at level. Therefore, regardless of the non-stationarity of the urbanization rate, there is no cointegration between the two time series.

Referring to the results in Table 1 the following equation is estimated before and after the threshold values for the regime

$$\:{CO}_{2}PC=\alpha\:+\:{\beta\:}_{1}{UR}_{t}+\:{\beta\:}_{2}REP+\:{\epsilon\:}_{t}\dots\:\dots\:..\left(1\right)$$

where

URt: urbanization rate/threshold repressor,

REP: share of renewables in electricity production (wind and solar),

B1: coefficient of urbanization rate before and after the threshold value,

B2: coefficient of the share of renewables in electricity production before and after the threshold value,

According to the threshold regression results in Table 4, the relationship between CO2 emissions per capita and urbanization rate is nonlinear. This is because the sign of the coefficients of the urbanization rate turned from positive in Regime 1 (1990–2003) to negative in Regime 2 (2004–2018) and then to positive in Regime 3 (2019–2022).

Table 4

Results of threshold effects

Variable | Coefficient | Std. Error | t-Statistics | Prob |

| \(\:\varvec{R}\varvec{e}\varvec{g}\varvec{i}\varvec{m}\varvec{e}\:1:\varvec{U}\varvec{R}<0.7831\:\:\:\:\:\:\:\:\:\:14\:obs\) | | | |

UR | 2.840 | 4.95 | 0.573 | 0.5714 |

REP | -0.003 | 0.008 | -0.469 | 0.6432 |

C | 0.819 | 3.85 | 0.212 | 0.833 |

| \(\:\varvec{R}\varvec{e}\varvec{g}\varvec{i}\varvec{m}\varvec{e}\:2:\:\varvec{O}.7831\le\:\varvec{U}\varvec{R}<0.9119\:\:15\:obs\) | | | |

UR* | -7.217 | 0.98 | -7.355 | 0.0000 |

REP* | -0.04 | 0.014 | -2.808 | 0.0097 |

C* | 9.334 | 0.832 | 11.207 | 0.0000 |

| \(\:\varvec{R}\varvec{e}\varvec{g}\varvec{i}\varvec{m}\varvec{e}\:3:\:0.9119\:<UR\:\:\:\:\:\:\:\:4\:obs\) | | | |

UR | 266.69 | 159.57 | 1.671 | 0.1077 |

REP | -0.0979 | 0.075 | -1.288 | 0.2098 |

C | -239.70 | 144.41 | -1.659 | 0.1100 |

| \(\:{R}^{2}=0.9038\) Adjusted R-Squared = 0.88717 | | | |

| F-statistics = 28.1948 Prob (F-statistics) = 0.00 | DW Stat = 2.36 | |

Source: Using Eviews 12 |

*Significant at 1% |

The threshold values of the urbanization rate are (0.78, and 0.91), which are consistent with the results of the sequential F-statistics determined threshold test shown in Table 1. Although the signs of the coefficient of the renewables in electricity production (REP) are negative in all regimes, these signs mean the REP controls the form of the relationship between urbanization rate and CO2 emissions per capita. The mechanism of this control is that if the REP is below a specific percentage value, then the relationship is positive between the urbanization rate and CO2 emission. By contrast, if the REP variable exceeds a specific percentage value, then the relationship will be negative. To support this view, the study carefully analyses the regression results in Regime 2 and compares these with the results in Regime 1.

In Table 4, the results of the threshold regression in regime 2 (2004–2018) show that the relationship between the independent variables urbanization rate and share of renewables in electricity production and the dependent variable CO2 emission per capita is a statistically significant negative relationship. This means that Jordan successfully reduced its CO2 emissions per capita despite the increasing urbanization rate. This success results from the country’s increasing dependency on renewable to generate electricity to cover the electricity demand caused by the increase in urbanization rate. A simple comparison between regimes 1 and 2 shows the compound growth rate of the share of renewables in electricity production for regime 2 (2004–2018), which grew by 24% while the compound growth rate of per capita electricity generation grew by 1.23%. While these compound growth rates in regime 1 (1990–2003) are very close, they reach 5% and 3.2% for the share of renewable in electricity production and per capita electricity generation, respectively. Depending on the results of threshold regression in regime 2, and as the urbanization rate in Jordan reaches approximately 0.91, the growth rate of renewable electricity production should be approximately 20 times bigger than the growth rate of electricity production from traditional resources to make an effect to decline CO2 emissions. Therefore, Jordan should attract investments and encourage research in the clean energy sector.

The final step is to estimate the economic impact. To examine the effect of urbanization rate and share of renewables in electricity production on gross domestic product per capita the study estimates the following equation

$$\:GDPPC={\alpha\:}_{0}+\:{\alpha\:}_{1}{UR}_{t}+\:{\alpha\:}_{2}REP+\:{\epsilon\:}_{t}\dots\:\dots\:..\left(2\right)$$

GDPPC: GDP per capita

URt: urbanization rate

REP: share of renewables in electricity production (wind and solar),

\(\:{\alpha\:}_{0}\) : intercept

\(\:{\alpha\:}_{1}\) : Coefficient of urbanization rate after the threshold value

\(\:{\alpha\:}_{2}\) : Coefficient of the share of renewables in electricity production before the threshold value

The OLS method and the period (2004–2022). The study choses this period due to this period includes the launching of the national energy strategy (2007–2020) and the current urbanization rate which is 0.91. According to the results in Table 5, the coefficients of urbanization rate and share of renewables in electricity production are positive and statistically significant at 10% level.

Table 5

Economic impact of urbanization rate and share of renewable in electricity production on GDP per capita

Variable | Coefficient | Std. Error | t-Statistics | Prob |

UR | 3493.208 | 1738.46 | 2.0093 | 0.0628* |

REP | 15.43 | 7.7707 | 1.9864 | 0.0656* |

C | 640.725 | 1497.44 | 0.4278 | 0.6748 |

| \(\:{R}^{2}=0.5709\) Adjusted R-Squared = 0.5137 | | | |

| F-statistics = 9.98 Prob (F-statistics) = 0.001752 | | | |

Source: Using Eviews 12 |

*Significant at 10% |

This means that as the urbanization rate and the share of renewable in electricity production increase, the GDP per capita increases too which is consistent with (Adem et al. 2020 and Sheng et al 2017).