Environmental Kuznets curve (EKC) across the globe
[INSERT TABLE 1 AROUND HERE]
We deduced that 26 countries confirmed the EKC hypothesis in terms of economic and statistical significance (Table 1). Similar to the results of prior studies, we partially confirmed the EKC hypothesis. Specifically, the EKC hypothesis was confirmed for central and eastern European countries51–54, while the results were mixed for western and northern European countries55,56, as well as Asian countries11,57–60.
Among these countries, 16 have crossed the turning point (decoupled countries), whereas 10 have not (non-decoupled countries). Tables 1a and 1b show the list of decoupled and non-decoupled countries, respectively. Interestingly, while most countries that crossed the turning point portrayed an inverted U-shaped relationship between the income per capita and carbon emissions, a few exhibited a gamma-shaped relationship.
[FIGURE 1 AROUND HERE]
Consistent with the EKC hypothesis, non-decoupled countries showed a positive relationship between the GDP per capita and the carbon emissions per capita (Fig. 1). Meanwhile, gamma-shaped and inverse U-shaped distributions were observed for decoupled countries. Supplementary Material 1 shows the data of a few countries that have not yet crossed the turning point. Brazil, Saudi Arabia, and the Republic of Korea are examples of countries that are non-decoupled, indicating that their economic growth and carbon emissions have a positive relationship.
The inverse U-shaped EKC indicates an ideal balance between economic development and emission levels. Notably, an inverse U-shaped relationship is ideal for climate-change-sensitive governments to achieve carbon neutrality as these governments aim to reduce carbon emissions, while maintaining positive economic growth. Supplementary Material 1 shows data of countries (e.g., Germany, France, New Zealand, Singapore, Sweden, and Switzerland) that portrayed an inverse-U-shaped relationship.
Simultaneously, we observed a gamma-shaped distribution relationship in the decoupled countries, i.e., these countries failed to reduce carbon emissions, even after they crossed the tipping point and their overall economy improved. While the absolute turning point value varied for each country, we observed a similar trend in most decoupled countries. Countries that seek carbon emission reduction prefer an inverse U-shaped relationship. The gamma-shaped relationship implies that economic development after a threshold is unrelated to the reduction in emissions levels. Supplementary Material 1 presents examples of countries that followed a gamma-shaped distribution, e.g., Australia, Canada, and Japan.
[FIGURE 2 AROUND HERE]
We further observed these patterns at the continental and regional levels by observing and determining whether the gamma- and inverse U-shaped relationships portrayed continent- or regional-level patterns. The EKC grouped by continent (including only important events) may affect carbon emissions (Fig. 2). Europe was the only continent in which an inverse U-shaped relationship between economic growth and carbon emissions was observed. Moreover, our results revealed that carbon emissions decrease after the GDP per capita crossed approximately USD 3,500 (USD value in 1960). Supplementary Material 2a shows a graphical representation of the EKC changes in Europe with respect to different periods (1960–1979, 1980–1999, and 2000–2019). Notably, while economic growth and carbon emissions followed a positive relationship in the 1960s, this trend flattened in the 1980s and subsequently declined sharply in the 2000s. Since then, important events have affected carbon emissions, such as the adoption of the European Union (EU) ETS Phase 1 (2005), Phase 2 (2008), and Phase 3 (2013) and the Paris Agreement (2015).
[TABLE 2 AROUND HERE]
Additionally, we conducted a univariate analysis to examine the effectiveness of the ETS and compared its effectiveness with other global events. We chose the first (1973) and second (1978) oil shocks as global events that may have significantly affected carbon emissions. This is because oil price fluctuated heavily during the crises, which in turn affected firm-level production and household energy usage. First, we measured and compared differences in the 3-year average emissions before and after the event year for each continent. Panel A and Panel B of Table 2 show the results for the first and second oil shocks, respectively. Carbon emissions before and after the event portrayed no significant difference except for Oceania (Panel A) and North America (Panel B). Panel C shows the results for the three phases of the EU ETS. Similar to the previous two tests, we calculated the 3-year average before and after the enactment of each ETS phase and checked the differences. Regardless of the ETS phase, we observed a significant decrease in carbon emissions. For instance, the carbon emissions per capita reduced by 3.1%, 9.4%, and 10.7% during EU ETS Phases 1, 2, 3, respectively. In terms of the absolute scale, the 3-year average carbon emissions before Phase 1 were 8.894 tons per capita. This reduced to 7.330 tons per capita in the 3-year average carbon emissions after the ETS Phase 3 was implemented.
Africa, Asia, Oceania, and South America were yet to reach the turning point and followed a positive relationship between economic growth and carbon emissions. An interesting picture emerged in North America as the carbon emissions decreased sharply at the GDP per capita of USD 1,000 and USD 2,000 (USD value in 1960) and remained stable thereafter. We analyzed the data for North America further for decades (Supplementary Material 2b) and found that carbon emissions declined sharply in 1960–1979, and subsequently increased slightly, yet consistently. In the 1970s, two oil crises directly affected the U.S.A. causing the manufacturing production to decline sharply, and the stock market to crash8,61.
ETS effects
The continent-level data portrayed the statistical significance of climate change policies and schemes; therefore, we further investigated their effects. Many countries that have tried to achieve similar goals have implemented various policies. If such policies are effective, then the countries that have implemented them would experience a decrease in carbon emissions compared with countries that have not implemented such policies yet. However, it is difficult to compare the policy effects empirically and statistically between countries. This is even more difficult when the policy content varies across countries. Therefore, we had to choose a regulation that was implemented by multiple countries as a reference policy to divide the country-level sample observations. Therefore, we analyzed the ETS implemented in multiple countries and investigated its effect on the relationship between the GDP per capita and carbon emissions per capita of every country. Further, we divided the samples before and after the turning point and observed the ETS effects. The underlying conjecture of our analysis was that, for countries that implemented the ETS before they crossed the turning point, the ETS would reduce the positive coefficient between the GDP per capita and the carbon emissions per capita. However, for countries that implemented the ETS after crossing the turning point, a more negative relationship would occur between the GDP per capita and carbon emissions per capita.
In this section, we explored the effects of ETS on the EKC for the decoupled, as well as non-decoupled, countries. Specifically, we considered the ETS as a benchmark policy shock for two reasons. First, the ETSs are aimed at regulating firm-level carbon emissions, which are often closely related to human activities, and these emissions are the primary drivers of climate change7,31,57,62. Therefore, studying regulations related to carbon emissions may be more helpful than studying those related to other air pollutants. Second, the ETS follows a market-based mechanism in which policymakers hope to effectively reduce emissions without compromising economic growth. In market-based environmental regulation policies, environmental protection may become a business activity, wherein the cost of emissions can be internalized and marketized63. Taking full advantage of the ETS, we designed both univariate and multivariate settings to explore the ETS effects on the relationship between carbon emissions and economic growth.
[FIGUIRE 3 AROUND HERE]
Decoupled countries started to adopt the ETS in the mid-2000s. After ETS adoption, carbon emissions per capita started decreasing significantly compared with the countries that did not adopt the ETS (Fig. 3a). Moreover, the decoupled countries that adopted the ETS showed lower carbon emissions. This implies they may have already invested in low-carbon technologies and production methods and employed the ETS to further accelerate emission reduction.
Theoretically, before decoupling occurs, the relationship between the GDP per capita and carbon emissions per capita should be positive64. This relationship was observed for non-decoupled countries (Fig. 3b) that did not adopt the ETS. However, the countries that adopted the ETS showed decreasing carbon emissions in the 2010s. Considering that countries began to adopt ETS in the mid-2000s, our results imply that ETS adoption has somehow affected the overall economy of the nation and decreased the emission per capita.
[FIGURE 4 AROUND HERE]
Figure 4 presents a more direct comparison of the carbon emission changes after ETS adoption. We averaged the carbon emission per capita of sample countries before and after the ETS adoption. The overall carbon emission per capita was reduced in decoupled samples, which is consistent with the findigns of a prior study64. Additionally, the reduction rate was higher for countries that adopted the ETS, implying that effective ETS strategies can help achieve sustainable development. Moreover, consistent with EKC theories, the overall carbon emissions per capita portrayed a significant increase in the countries that did not adopt the ETS. Notably, the carbon emissions of the countries that adopted the ETS increased at a lower rate. Overall, the results implied the significance of ETS in developed and developing countries. However, as developing countries will inevitably emit more carbon while achieving higher productivity, ETS may have slowed the increment rate of carbon emissions in the non-decoupled countries that did not adopt the ETS.
[INSERT TABLE 3 AROUND HERE]
Based on the univariate results, we conducted multivariate regression models to test the ETS effects. Specifically, we used the sample countries for which the EKC condition was applicable. Further, we divided the samples into pre-industrial and post-industrial economies and tested the ETS effects. The empirical results confirmed that ETS adoption, before and after the turning points, had a positive effect on carbon neutrality. The results were consistent, regardless of the model used. Table 3 shows the results of the regression analysis. Our results revealed that ETS adoption after the turning point further increased the negative relationship between the GDP per capita and the carbon emissions per capita (Column 1 and 2). Moreover, ETS adoption reduced the positive association between economic development and carbon emissions before the turning point (Column 3 and 4). Overall, the estimation results imply that ETS adoption in an early stage of economic development is an effective strategy to reduce the emissions per capita, while ETS adoption after the turning point is also effective in reducing carbon emissions.