This theory, initially introduced by Simon Kuznets in 1955, describes the inverted U-shaped relationship between economic growth and income inequality. Although it did not originally address environmental issues, Grossman and Krueger ³¹ later applied this concept to the relationship between economic growth and environmental degradation (ED), identifying a similar pattern. The curve indicates that ED tends to rise during the early stages of economic growth but declines as growth continues. ³² This relationship can be understood from two perspectives: the two phases of early and later economic development or the three stages of pre-industrial, industrial, and post-industrial economies. ³³

Based on the Figure 2 presented below, development at each stage is influenced by shifts in production capital, including labor and human capital, in addition to technological innovation. Technology, as an endogenous variable, relies on investments in research and human capital, as outlined in Endogenous Growth Theory. ³⁴ These dynamics facilitate the transition from a negative to positive relationship between economic growth and ED or its substitution variables. ³⁵

From various empirical studies, the EKC has been widely used by environmentally oriented researchers and continues to evolve the use of indicators to analyze causality between economic and environmental indicators. However, most often analyzed the relationship between GDP and CO ₂ and used approximately 100 articles out of 200 collected from their paper. ³³ In recent years, various indicators have been used to empirically analyze the existence of the EKC, such as energy consumption (renewable and non-renewable energy) and GDP. ^{37– 39}

It should be noted that there is still a gap in the selection of environmental indicators, as not all indicators can be used to show an inverted U-shaped relationship due to the lack of consistent data. This is based on empirical results, and researchers have found that for industrial pollution and human health, it has not been possible to assess the EKC. ⁷ Furthermore, literature that comprehensively examines the existence of the EKC in Eurasia from the geographical area of 93 countries still does not exist. On average, researchers focus their research in selecting only a few countries ⁴⁰ or within the scope of smaller regions such as the EU countries ^{41, 42} and focus on continents such as Eastern Europe ⁴³ and Central Asia ¹⁶ respectively, which proves the existence of a valid EKC.

Human development is a central goal for nations striving to achieve prosperity, aligned with Sustainable Development Goals (SDGs). ^{44, 45} It focuses on creating an environment that enhances human capabilities, allowing individuals to lead longer, healthier, and fulfilling lives. This concept addresses the limitations of traditional economic growth, in which a high GDP does not necessarily translate into comprehensive development. ⁴⁶ To measure progress, the Human Development Index (HDI) was developed to, assess long and healthy lives, education, and standard of living as indicators of development and welfare.

The educational dimension of the HDI is crucial in shaping environmental outcomes. Education fosters ethical awareness and responsibility to protect the environment and prioritize sustainability. ^{47, 48} A study conducted in OECD countries shows that improvement in human development enhances environmental sustainability, reducing CO ₂, greenhouse gases, and air pollution. ⁴⁹ However, the relationship between HDI and environmental degradation (ED) varies. For instance, Li and Xu ²⁴ observed an inverted U-shaped relationship in China but noted a positive correlation between HDI and specific ED aspects, such as waste gas and industrial solid waste. Similar findings were reported by Rahayu and Handri ⁵⁰ in Indonesia and Polat and Çil ²³ in eco-innovative countries, where the HDI exhibited an EKC pattern with ED. Conversely, a study had conducted on 114 countries with a division of 28 developed countries and 86 developing countries, show that the increase in HDI in developed countries has a positive and significant effect on the Environmental Development Index (EPI), but contrasts with the results in developing countries where the effect is not significant. ⁵¹

Some studies also report a negative relationship between HDI and environmental quality, attributing it to countries that prioritize conventional economic growth over sustainable practices despite achieving higher levels of development. ^{19, 52} This study explores the complex interplay between the HDI and ED across developed and developing Eurasian countries, offering a novel regional perspective.

Democracy is a system of government that upholds natural rights by allowing citizens to participate in the creation of laws and preserving freedoms such as the right to happiness, life, and speech. ^{53, 54} While democracy regulates these freedoms to prevent conflict, it also emphasizes citizen involvement in lawmaking. ⁵⁵ Democratic freedom includes freedom of speech, religion, freedom from want, and freedom from fear, as outlined by President Roosevelt in 1941. However, the implementation of democracy varies across countries, shaped by cultural, constitutional, economic, and ideological factors. To measure democracy levels, this study uses the Democracy Index from the Economist Intelligence Unit (EIU), which categorizes countries as full democracies, flawed democracies, hybrid regimes, or authoritarian regimes, based on five indicators: electoral processes, government functioning, political participation, political culture, and civil liberties.

Democracy significantly contributes to environmental quality by involving the public in shaping policies, ensuring transparency, and upholding accountability. ⁵⁶ Democratic systems provide citizens with a platform to voice concerns, including environmental issues, that can mitigate environmental degradation. ²⁹ Studies confirm this relationship, such as Zandi and Abidin ⁵⁷ and Zeinalzadeh, ⁵⁸ who find that increasing democracy reduces environmental degradation in ASEAN and OIC countries. Moreover, Lindvall and Karlsson ⁵⁹ reviewed 72 studies and concluded that democracies produce better climate policies than autocracies do. Democracy positively influences environmental quality through political rights, information transparency, electoral accountability, and active environmental interest groups, fostering greater public awareness. ^{29, 60, 61}

However, some studies have argued that democracy can negatively impact environmental degradation. Frequent electoral changes can disrupt long-term environmental strategies, creating uncertainty and weakening regulations. ^{62, 63} A study shows that the quality of democracy matters, noting that democracies with low corruption indices are less effective in reducing CO ₂ emissions than those with higher corruption indices. ⁶⁴ Socioeconomic challenges in developing countries often deprioritize environmental issues. ⁶⁵ Thus, the impact of democracy on environmental quality is influenced by institutional strength and socio-political contexts.

This study uses the EKC hypothesis to explains the inverted U-shaped relationship between economic growth and environmental degradation. The mathematical function used refers to the research of Jan and Sheikh. ⁶⁶ This function is written as: E it = f ( Y it , Z it , F i , F t , ε it ) (1)

From the function above, E denotes Emission or CO ₂ specifically, i for cross-section, which in this study is the country with the value of i = 1,2,…… N, and t denotes time series in years with the value of t = 1,2,……. T. Y describes the income of the country, Z denotes other variables or factors that can affect the EKC hypothesis formula, or CO ₂ emissions as the dependent variable. F _i denotes the cross-sectional effect, F _t denotes the time-series effect, and ε _it denotes the error term of the model. Then, to determine the effect of economic growth on environmental degradation, as stated by Stern, ⁶⁷ a quadratic equation of economic growth was applied. The result will be inverted U-shaped if the results of the quadratic equation are β1 > 0 and β2 < 0. ³² The equation for the fundamental EKC hypothesis is as follows: E it = α i + θ t + β 1 Y it + β 2 Y it 2 + β 3 Z it + ε it (2)

Where α denotes country-specific intercepts and θ denotes time-specific intercepts. This equation is the standard form presented by Fredriksson and Neumayer. ⁶⁸ After obtaining the basic equation, we added the research variables used. CO ₂ emissions were taken as a proxy for environmental degradation, which is the dependent variable. Other independent variables included economic growth, human development, and the quality of a country's institutions. The selection of these variables is, of course, full of consideration and follows previous empirical evidence that results in an influence on E _it. Thus, the empirical model is as follows lCO 2 it = α i + θ t + β 1 lGDP it + β 2 lGDP it 2 + β 3 lDM it + β 4 lHDI it + ε it (3)

From this last equation, where CO ₂ is known as the denotation for carbon dioxide and l is log-natural, we added it to all our variables to increase the efficiency of estimation results, unify units with elasticity of interpretation, and overcome normality. ⁶⁹ Furthermore, GDP is denoted as economic growth, with GDP ² as the quadratic equation in determining whether there is an inverted U-shaped relationship with CO ₂. Then, there is DM, which is intended for democracy variables that are incorporated from five indicators, as described in the literature review section. Finally, HDI was used for human development.

In summary, we have summarized each variable, from its definition and proxy to the secondary data sources we used, presented in Table 1 below. We use fossil CO ₂ emissions metrics, which include the measurement of all types of emissions from fossil fuel consumption (e.g., oil, coal, flaring, gas, cement, and bunker fuel), as described by Friedlingstein. ⁷⁰ We used fossil CO ₂ to represent the dependent variable of environmental degradation from various environmental dimensions, although it is more concentrated on direct air pollution. ³ The source we use for this variable is the Global Carbon Budget, which is both related to methodological information and databases.

For GDP, we use constant GDP with 2015 as the base year and US$ currency. As the focus in seeing the real growth of a country is better by looking at the quantity of goods and services and fulfilling consumption needs without being affected by price fluctuations or inflation, a constant GDP is the right choice. ⁷¹ We choose constant GDP as a proxy for economic growth, with a focus on the industrial aspect of CO ₂ environmental degradation. The source of this constant GDP data is the World Bank.

Furthermore, the Human Development Index variable is used as a proxy for human development in a country. Our variable is used to measure an individual's quality of life, which cannot be described by GDP per capita alone or by other monetary approaches. The measurement of the HDI is dimensionally used by all countries in the world, namely, long and healthy life, education, and standard of living, but the constituent indicators and the weight of each indicator can differ from one country to another according to the circumstances of the problems faced. ²⁰ The source was used to obtain the data.

Finally, we use the Democracy Index variable to represent the quality of a country’s government. This variable consists of five indicators of democracy: electoral process and pluralism, government functioning, political participation, political culture, and civil liberties. The index calculation that has been combined from the five indicators classifies countries in the world based on four groups: full democracy, flawless democracy, hybrid regime, and authoritarian regime. The data for this Democracy Index were sourced from the EIU.

In this study, we used panel data from selected 30 countries in Eurasia for a range of years between 2015 and 2022. Statistically, this equals the sum of ( N = 30) and ( T = 8). A purposive sampling technique was used in this study to determine whether the countries in Eurasia are too large, or too small, or have significant internal problems, such as wars and politics, we selectively consider them to avoid error factors that interfere with the results of the model's influence. The identification of the variables tends to be dynamic from year to year, which means that the variables can be influenced by lag variables or the previous time. Therefore, dynamic panel data regression is appropriate as an estimation method tool with the following model: y it = α i + β ′ it + yy it − 1 + u it + ε it (4)

The Generalized Method of Moments (GMM) approach proposed by Arellano and Bond was utilized to address endogeneity and inconsistency issues, with an emphasis on the System GMM (Sys-GMM) method developed by Blundell and Bond. The Sys-GMM combines first-difference and level equations to improve instrument strength and effectively handle reverse causality.

This study uses the two-step Sys-GMM approach and adds instrument variables (iv), which is a robust method consistent with prior empirical studies. ⁷² The Hansen-J and Arellano-Bond tests were conducted to ensure model validity by checking for over-identifying restrictions and no second-order serial correlation (AR(2)). Additionally, Pesaran’s CD-Test was used to examine cross-sectional dependence across panel groups as the panel setup met the N > T criterion. ⁷³

Considering the persistent nature of atmospheric CO ₂ and its long-term impact, we treat it as an endogenous variable and include lags up to three periods (1–3 years) in the model, following. ⁷⁴ While adding lags improves the model's dynamics, it may reduce the explanatory power of the other independent variables, representing a trade-off. Expanding the sample size in future research could address this limitation. ⁷⁵

Before estimating the model, we present the descriptive statistics of the research variables in Table 2. This includes the total number of observations (OBS), standard deviation (SD), and minimum (Min) and maximum (Max) values for each variable, location, and year. The significance level (α) was set at 5%, where a hypothesis was considered significant if the p-value was below this threshold ( H ₁) and insignificant if it was above ( H ₀). However, for the EKC hypothesis, the significance criteria differ, as explained in the econometric model section.

For cross-sectional dependence (CD-Test), we reject the null hypothesis ( H ₀) for variables lCO ₂, lGDP, lGDP2, and lHDI but accept it for lDEM. Descriptive statistics indicate 240 balanced panel observations, ensuring no variation in the sample size across the variables. A multicollinearity test was conducted on all the independent variables, as shown in Table 3. The correlation values for each variable remained within the acceptable ranges. While the correlation between GDP and GDP ² is relatively high, this is typical in traditional EKC tests. ⁷⁶ If multicollinearity is present, long- and short-term elasticity coefficients could be used as an alternative method, as suggested by Narayan. ⁷⁷ However, as indicated by Zainodin and Yap, ⁷⁸ correlation values below 0.75 confirm that multicollinearity is not a concern in this study.

First, we assessed potential endogeneity factors in the model using the Hansen-J test to evaluate instrument validity and ensure no over-identifying restrictions or heteroscedasticity issues. The Arellano-Bond test was then applied to confirm the absence of second-order serial correlation in errors (AR2). As shown in Table 4, the Hansen-J test results for lags 1-3 consistently fall between 0.05 and 0.80, indicating optimal model validity, and the null hypothesis (H0) is accepted. ⁷⁹ Similarly, the AR2 test results for lags 1-3 exceed the threshold of 0.05, supporting the absence of serial correlation and the acceptance of H0.

All independent variables are treated as exogenous with appropriate instrumental variables and lagged explanatory variables consistently address endogeneity issues. ⁸⁰ Table 4 presents the estimation results for lag 3. However, the variance in errors remains uncontrolled, owing to the diverse characteristics of the countries studied. This may stem from imbalances and unique developmental differences between developing and developed countries in Asia and Europe, as outlined in the Background section. Thus, the findings represent a general overview of the Eurasian region, without distinguishing the effects specific to developed or developing countries.

The information that can be utilized from Table 4 is that the lag CO ₂ at all levels is significant at 1% for all columns. Further, the coefficient for GDP ² is consistently negative and significant across columns, with positive and significant GDP values for Models 1 and 3. We can confirm that, in general, countries in Eurasia have achieved an inverted U-shaped EKC relationship between economic growth and CO ₂, meaning that there is a decoupling process in the region. This finding supports the results of previous empirical studies that prove the existence of the EKC between economic growth and CO ₂ in Eurasian countries, Europe, and Asia in general. ^{41, 43}

Then, the HDI variable is consistently negative and significant at the 1% level in all columns and lag levels. This means that increasing human development in Eurasian countries tends to reduce. This finding is supported by research from. ^{49, 50, 81} Democracy also consistently has a negative and significant effect, at least at the 5% level of the reasonable limit set by this study. This shows that countries with higher levels of democracy tend to suppress CO ₂ levels in Eurasia in general, which is also supported by previous empirical studies. ^{57– 59}

As previously explained, we conducted separate analyses for developing and developed countries in Eurasia using balanced total samples to minimize errors and provide clearer insights, as shown in Table 5. The estimates confirm a clearer direction of influence, with GDP having a positive significant effect, and its quadratic term showing a negative and significant effect. This confirms the existence of the EKC in both groups, suggesting progress towards environmentally friendly industries and the growth of the service sector. The reduction in CO ₂ emissions was more pronounced in developed countries, as indicated by the larger coefficients. The literature supports this trend, attributing it to ambitious CO ₂ reduction efforts through sustainable energy transitions and the gradual phase-out of conventional industries. ⁸² These efforts highlight the need for technology and knowledge transfer from developed to developing countries to facilitate energy transitions, as suggested by Kohnert, ⁸³ who emphasized similar solutions for sub-Saharan Africa. This finding aligns with previous research indicating that developing countries are more vulnerable to environmental degradation owing to disparities in development and infrastructure. ^{15, 84}

Furthermore, the Human Development Index section provides a consistent but contrasting relationship between developing and developed countries in Eurasia, with significant results at the 1% level and at the level of three CO ₂ lags. This shows that an increase in the HDI in developing Eurasian countries triggers an increase in the CO ₂ trend. This is in contrast to the situation in developed countries, where the increase in HDI is consistent in a situation that can reduce CO ₂. First, on average, developing countries are not yet inclusively using green energy for households, especially in Asia and SSA. ⁸⁵ Thus, their limited level of technology and industrial structure will increase their carbon footprint as HDI rises, with improved income, infrastructure, and consumption. ⁸⁶ For instance, Kuwait and Romania, despite having HDI values above 0.8, continue to show rising CO ₂ emissions due to inefficient industrial structures reliant on fossil energy. ⁸⁷ Conversely, developed Eurasian countries, with HDI values ranging from 0.8 to 0.9, demonstrate a decoupling of CO ₂ emissions from human development. This trend is supported by established infrastructure and increased environmental awareness, particularly through education, as shown in previous study. ²³

The last estimation analysis on the democracy variable consistently has a negative effect on CO ₂ emissions in both developed and developing countries in Eurasia and is significant at the 1% level. This can be confirmed based on the estimation results that the more democratic a country is, the much better is the reduction of CO ₂ levels than countries with authoritarian characteristics. ⁵⁶ However, this result is contrary to, ⁸⁸ who examined 37 OECD Countries. Furthermore, our findings that democracy decreases CO ₂ is more pronounced in developed countries than in developing countries, as seen from the coefficient value. This is acceptable given that the classification of full democracy with a score of 8.0 – 10.0 tends to be in developed Eurasian countries. ²⁶