The Effectiveness of Redistribution in Carbon Inequality: What About the Top 1%?

Abstract

This study investigates the impact of income redistribution on carbon emissions across 154 countries from 1995 to 2023, with a particular focus on carbon inequality. Using a dynamic panel approach with two-step System GMM estimations, the analysis considers three dependent variables: average per capita emissions, top 1% per capita emissions, and the ratio of top 1% per capita emissions to national average per capita emissions. Results show that income redistribution (measured in both absolute and relative terms) significantly reduces average per capita emissions in the short term. However, redistribution has no mitigating effect on the carbon emissions of the top 1%; in some models, it is even associated with increases in elite emissions and a widening of carbon inequality. These findings suggest that while redistribution may contribute to national emission reductions, it is insufficient to curb the carbon-intensive lifestyles of the wealthiest. The analysis confirms the Environmental Kuznets Curve (EKC) hypothesis and underscores the need for complementary policy tools to more effectively address the emissions of high-emitting individuals. Overall, this study contributes to the literature by linking income redistribution with emission disparities across income groups and highlights the importance of considering distributional dynamics in climate policy design.

1. Introduction

The opportunity costs of economic activities have gained increasing importance in the context of environmentally sensitive growth discussions. Whether based on production or consumption, economic growth inevitably generates environmental impacts. The nature and extent of these impacts are largely shaped by the level of income and, more critically, how this income is distributed within society. In this regard, one of the increasingly debated topics in environmental economics is the effect of income distribution on environmental outcomes. Among these, carbon emissions stand out as a key indicator when assessing environmentally conscious policies. Within this framework, a critical question arises: Does income distribution affect carbon emissions through economic activities?

Various studies addressing this question have examined the relationship between the level of development, the state of income distribution, and carbon emissions. However, the primary focus of this study is to investigate the impact of government policies aimed at redistributing income on carbon emissions. One of the two main features that will make this study distinctive and contributory lies here. There is literature investigating the impact of income distribution on carbon emissions. However, no study has been found that examines and predicts the impact of income redistribution in this breadth and detail. Specifically, this study will evaluate the potential impact of state interventions that reduce income inequality on environmental sensitivity and carbon emissions. Additionally, the second distinguishing feature of the study is the consideration of differences in income groups, such as the sensitivity of the top 1% of the wealthiest individuals in countries to carbon emissions. Based on this, testing the effectiveness of income redistribution policies will also constitute one of the main objectives of the study.

Roser et al. (2013) states that, following the Industrial Revolution, real income per capita increased approximately 14-fold, rising from USD 1102 in 1820 to USD 15,212 in 2018 [1]. During the same period, Ritchie et al. (2020) reported that annual carbon dioxide (CO₂) emissions increased more than 700-fold, from 50.7 million tons in 1820 to 37.12 billion tons in 2021 [2]. These figures indicate that one of the most significant opportunity costs of economic growth is carbon emissions and the accompanying environmental damage.

The rise in carbon emissions leads to various consequences, including climate change, extreme weather events, depletion of natural resources, scarcity, social unrest, and associated economic losses [3,4]. In addition, such environmental degradation poses serious threats to human health [5,6].

Historically, the shift to mass production and the associated increase in resource use have contributed to economic prosperity. However, these developments have severely damaged the ecological systems’ regenerative capacity [7]. Moreover, economic growth has also led to disparities in income distribution, commonly referred to as income inequality, which in turn has influenced environmental development. The economic consequences of income inequality have been widely discussed in the literature, with Kuznets being one of the pioneers in this field [8]. For instance, Hailemariam et al. (2020) explore the detrimental effects of income inequality on environmental quality [9].

This study aims to assess the impact of income inequality on environmental outcomes, with a particular focus on how income redistribution policies influence carbon emissions. In doing so, it also contributes to the growing literature on inequality and climate change. It seeks to understand how income disparities influence environmental sensitivity, particularly, how individuals in the top income percentiles differ from others in their environmental impact. The study also aims to explore whether government interventions can balance these differences in environmental sensitivity and thereby help reduce carbon emissions. These questions constitute the core research themes of this study.

2. Literature Review

2.1. Theoretical Background

In this section of the research, we discuss the theoretical basis and studies that examine the relationships between carbon emissions, income inequality, and redistribution. Initially, we explore the Environmental Kuznets Curve (EKC), which investigates the connection between income inequality and environmental degradation. This theory offers insights into the complex socio-economic forces driving environmental degradation and includes information on the variables selected for analysis. Subsequently, we review literature studies on the relationship between income inequality and redistribution.

2.2. Environmental Kuznets Curve (EKC)

The research draws its theoretical framework within the boundaries of the Environmental Kuznets Curve (EKC). Before the EKC, Kuznets (1955) pointed out economic growth and income distribution in his study and substantiated this explanation with the concept of an inverted U [8]. According to this curve, economic growth initially increases the incomes of skilled workers. This relationship between per capita national income and income inequality will empirically take the form of an inverted U. In Figure 1, the change in income inequality and per capita national income is visually represented.

Since the 1990s, when globalization gained momentum, the effects of economic growth on the environment have led to an increase in studies in the literature. Researchers, scientists, environmentalists, and policymakers have shown increasing interest in identifying the factors contributing to ecological degradation. The EKC examines the connection between income and environmental pollution. Although developed countries are generally held responsible for environmental degradation, developing countries also contribute to the problem by consuming natural resources and causing water, soil, and air pollution with industrial waste, often without realizing it [11]. Grossman and Krueger (1991, 1995) suggested that as a country’s economy grows, pollution levels first increase and then decrease, which is known as the EKC hypothesis [12,13]. However, a consensus on the EKC hypothesis has not been reached because it does not take into account the differences in energy consumption behavior and carbon emission intensity between low- and high-income groups. Figure 2 displays the EKC.

The basis of the analysis regarding the relationship between income inequality, economic development, and environmental degradation, and the role it plays, is the famous EKC hypothesis. The EKC hypothesis suggests, as shown in Figure 2, that there is an inverted U-shaped relationship between economic development and environmental degradation. That is, environmental quality deteriorates as per capita income increases in the initial stages of development but improves after a certain threshold of per capita income [9].

In pre-industrial societies, economic activities are based on agriculture, leading to no pollution, and they are not negatively affected by the relatively low per capita national income compared to the post-industrial period. The EKC has primarily gained validity with industrialization. With industrialization, the increased use of natural resources, emissions, technological developments, excessive emphasis on production outputs, and the neglect of the environmental impacts of growth have led to environmental degradation. However, as economic growth continues and life expectancy increases, cleaner water, improved air quality, and generally a cleaner habitat have become more valuable as people make marginal choices about how to spend their income [10].

According to Barbier (1997) [14], the EKC argues that there is an inverted U-shaped relationship between indicators of environmental pollution or resource consumption and per capita income level. That is, environmental degradation will initially increase but will eventually decrease as per capita income rises [14]. The subject’s development is based on research by Grossman and Krueger (1995) on the measurement of air quality [13]. They conducted extensive research on the claim that the economic growth brought about by the North American Free Trade Agreement (NAFTA) would promote environmental degradation. Shafik’s (1994) early EKC analyses present similar findings [15]. Panayotou (1995), in the analysis conducted for the International Labour Office’s World Employment Programme in 1992, perhaps explains the Kuznets-type U-shaped relationship between the rate of environmental degradation and the level of economic development in the earliest and most detailed manner possible [16].

Although there is empirical evidence supporting the EKC, it can be said that the theoretical framework of the Environmental Kuznets Curve is in the maturation stage. Lopez (1994) developed a theoretical model showing that under certain conditions, the relationship between pollution and income takes an inverted U shape with the liberalization of trade [17]. Munasinghe (1999) presents both a theoretical model and findings from empirical studies [18]. In his research on the marginal benefits and costs of reducing pollution, Munasinghe (1999) concluded that during early growth stages, decision-makers find the perceived benefits of environmental protection too small to justify sacrificing further economic development [18]. Cole et al. (1997) examined the relationship between per capita income and various environmental indicators using panel data across countries [19]. Richmond and Kaufmann (2006) analyzed the impact of fuel mix and development level on the existence and magnitude of a turning point in the relationship between income and energy use and carbon emissions [20]. After these studies, models targeting single or multiple countries with direct foreign investments, economic, demographic, and political variables were developed [21,22].

2.3. The Relationship Between Income Inequality and Carbon Emissions

Income distribution plays a crucial role in total carbon dioxide emissions and, consequently, global warming. Income inequality within a country’s population can affect carbon emissions through different energy consumption behaviors and patterns among income groups. Additionally, economic growth is generally thought to be accompanied by higher emissions. Ravallion (2000) suggested that the relationship between income inequality and carbon emissions depends on the marginal emission tendencies of different income groups, considering their consumption and emission behaviors [23]. For example, when the welfare levels of low-income groups increase, their tendency to consume environmentally harmful and low-cost energy sources like coal can lead to higher carbon emissions. It is thought that high-income individuals, who travel more and do not use public transportation, can also increase carbon emissions based on their consumption behaviors.

Recent studies conducted in the literature present findings on the existence of a contradictory relationship between income inequality and carbon emissions. Some of the studies interpret that income inequality and environmental impacts are inversely proportional. Additionally, Guivarch et al. (2021) suggest that the lowest income group, which has the least impact on greenhouse gas emissions, is the most affected by environmental issues [24]. Baiocchi et al. (2010) argue that the lifestyle of very wealthy households is significantly more carbon-intensive compared to the lifestyle of less wealthy households [25]. Chancel et al. (2023) argue that the main responsibility for global carbon emissions lies with the wealthiest 1% of the population; this income group is responsible for nearly a quarter of the increase in carbon emissions over the past 30 years [26].

The relationship between income inequality and carbon emissions becomes complex when considering the effects of income changes. As income inequality decreases, the increase in income for low-income groups often leads to higher consumption and consequently higher emissions, which is consistent with Keynes’s (1936) Absolute Income Hypothesis [27]. The increase in the welfare of low-income individuals leads to a rise in the demand for goods and services with environmental costs. A decrease in the income of high earners may lead to a reduction in their consumption. In this context, effective public policies should strive to raise the living standards of low-income individuals without causing carbon emissions [28]. Ravallion (2000) analyzed 42 country between 1975 and 1992 and concluded that high income inequality leads to low carbon emissions [23]. Grunewald et al. (2017) found that in low- and middle-income economies, higher income inequality is associated with lower carbon emissions, whereas in upper-middle-income and high-income economies, higher income inequality increases per capita emissions between 1980 and 2008 [29]. Liu et al. (2019) examined the relationship between income inequality and carbon emissions in the United States from both short-term and long-term perspectives [30]. According to the findings, high income inequality increases carbon emissions in the short term but reduces them in the long term. Chen et al. (2020) [31] stated that income inequality hardly affects carbon emissions in most developed countries. The model was applied to country-level data for 17 G20 countries from 1988 to 2015 [31]. Wang and Qu (2024) stated in their research conducted with a wide group of countries (193 countries) between 1990 and 2019 that the reduction of income inequality in low-income countries would increase carbon emissions [32].

Some studies have found that reducing income inequality can decrease carbon emissions and that there is a positive relationship between these variables. Golley and Meng (2012) conducted a study using the Urban Household Income and Expenditure Survey 2005 in urban areas in China, finding that high-income households cause more emissions compared to low-income households, indicating a positive relationship between income and carbon emissions [33]. Zhu et al. (2018) show in their study of BRICS countries from 1994 to 2013 that income inequality significantly and positively impacts carbon emissions in medium- and high-emission countries [34]. Baloch et al. (2020) found in their studies of 40 Sub-Saharan African countries between 2010 and 2016 that the rise in income inequality increased carbon emissions [35]. Khan et al. (2025) stated in their research conducted in Bangladesh between 1980 and 2021 that the increase in income among the highest earners would lead to an increase in carbon emissions [36]. In sum, the literature about the relationship between income inequality and carbon emissions varies as the models, years, datasets, and samples differentiate.

2.4. The Relationship Between Redistribution and Carbon Emissions

The primary focus of this study is the redistribution of income. Redistribution refers to the reallocation of income generated within a particular economic area after its initial distribution based on ownership of production factors [37]. This process aims to reduce inequalities and differences in quality of life. The key instruments in such policies are progressive taxation and social benefits, which create an impact based on their levels and progressivity [38].

This study examines the impact of income redistribution policies on carbon emissions. The research question explores whether income redistribution leads to a reduction in carbon emissions. Existing studies on this topic provide mixed findings, mentioning both positive and negative effects.

Andersson (2024) states that efforts to reduce economic inequality have shown limited effectiveness in mitigating environmental impact [39]. Similarly, Coşkun (2025) argues that policies aimed at reducing income inequality tend to increase both consumption-based and production-based carbon emissions due to their emphasis on economic growth rather than environmental sustainability [28]. Conversely, Andersson (2023) suggests that income redistribution efforts have a positive impact on reducing carbon emissions [40]. In line with this, Jorgenson et al. (2025) identify a negative correlation between reducing income inequality and carbon emissions [41]. Sahu and Mahalik (2025) further highlight that reducing income inequality could increase demand for renewable energy, emphasizing the need for policymakers to enhance intergenerational mobility to create a positive impact [42].

A range of environmental policies are implemented to reduce carbon emissions, including carbon taxation. A carbon tax is a regulatory measure designed to prevent negative externalities. However, from an income redistribution perspective, it may have adverse effects. Just as the effectiveness of every policy needs to be analyzed, the economic and social effectiveness of carbon taxation should also be assessed. According to Pranita and Sarjana (2025), the burden of carbon taxation may disproportionately impact low-income individuals, as a significant portion of their income is allocated to energy expenses [43]. Beyond determining the tax rate, various expenditure-side measures such as energy subsidies, direct cash transfers, or tax refunds are essential tools in addressing carbon emissions through income redistribution [43].

Income redistribution policies play a critical role in mitigating the adverse effects of income inequality. Addressing environmental externalities, particularly carbon emissions, requires targeted interventions. When income inequality negatively impacts carbon emissions, income redistribution policies become an essential mechanism for counteracting these effects. Ridzuan (2024) emphasizes the role of government support in housing and transportation for low-income groups, as these are areas where their relative expenditures are high [44]. Considering the significant carbon emissions associated with these sectors, the following income redistribution-driven measures are deemed necessary and rational for reducing emissions [44]:

• Housing: encouraging electrification for heating and other household needs, reducing dependence on liquefied petroleum gas, and implementing insulation policies to improve energy efficiency.
• Transportation: expanding investments in electric-powered public transportation and railway freight systems, introducing tax incentives for electric and zero-emission vehicles, and developing charging infrastructure.

3. Data, Sample, Methodology, and Models

3.1. Research Data and Sample

We constructed panel data for 154 countries worldwide using annual observations spanning from 1995 to 2023. The list of the 154 countries included in the empirical analysis is provided in Appendix A. The dependent variables include the per capita carbon emissions of the top 1% income group across countries; the overall per capita carbon emissions at the country level; and the ratio between these two, referred to as carbon inequality. Data on the per capita carbon emissions of the top 1% were retrieved from the World Inequality Database (WID) [45]. Carbon emissions per capita were obtained from the Our World in Data (OWID) database [46,47,48]. Gross Domestic Product (GDP) per capita and population, both included as explanatory variables, were also sourced from the same database.

The cross-country Gini coefficient, also included as a control variable, was obtained from the Standardized World Income Inequality Database (SWIID), which measures inequality based on households’ after-tax disposable income [49]. Descriptive statistics of the selected variables are presented in

Table 1. Descriptive statistics of variables.

Indicator	Definition	Source	n	Mean	SD	Median	Min	Max
$lnCO2$(1)	Per capita carbon emissions of the top 1% income group (in natural logarithm)	WID	4048	3.602	1.137	3.685	−1.529	6.433
$lnCO2$(2)	Carbon emissions per capita (in natural logarithm)	OWID	5411	0.607	1.572	0.867	−3.816	4.338
$CO2inequal$	Ratio of top 1% to average carbon emissions	Calculated	4075	32.929	62.036	16.783	−27.699	1201.194
$lngdp$	GDP per capita (in natural logarithm)	OWID	4480	8.994	1.228	9.121	5.889	12.004
pop	Total population (in billions)	OWID	5452	3.66	1.36 × 10⁸	7,483,800	9299	1.44 × 10⁹
gini	Gini coefficient	SWIID	4281	38.678	8.088	38.4	22	65.2
lnredist (1)	Absolute income redistribution (in natural logarithm)	SWIID	4072	1.556	1.000	1.280	−2.302	3.250
lnredist (2)	Relative income redistribution (in natural logarithm)	SWIID	4072	2.345	0.964	2.013	−1.469	3.969

.

The Gini coefficient, originally developed by Corrado Gini (1884–1965), is among the most widely used indicators of income inequality due to its intuitive interpretation and computational simplicity [50]. It ranges from 0 (perfect equality) to 1 (maximum inequality) and allows for the evaluation of redistributive policy effectiveness by comparing pre-tax and post-tax income distributions [51]. Building on this widely accepted framework, the present study employs standardized measures of income inequality and redistribution from the SWIID database [49]. SWIID harmonizes data from diverse international sources to provide comparable estimates of market and disposable income inequality across a broad set of countries and years. These features make it particularly suitable for large-n panel studies. Moreover, SWIID-based Gini indicators and redistribution metrics have been extensively utilized in empirical research examining the inequality–emissions nexus [9,29,31], reinforcing their methodological reliability and relevance for cross-country analysis.

Following the SWIID methodology, we measure the independent variable, income redistribution, in two ways: (1) absolute redistribution, calculated as the difference between market-income inequality and net-income inequality, and (2) relative redistribution, defined as the proportion of inequality reduced through taxes and transfers, calculated by dividing absolute redistribution by market-income inequality.

3.2. Methodology and Models

In this study, analyses were conducted based on three different dependent variables related to carbon emissions: (1) the overall per capita carbon emissions at the national level; (2) the per capita carbon emissions of the top 1% income group across countries; and (3) the ratio of the top 1%’s per capita carbon emissions to the national average, referred to as carbon inequality. The primary objective is to assess whether income redistribution—measured via absolute and relative metrics—has a significant and robust impact on these emission indicators while controlling for a set of socio-economic control variables, such as income inequality, population, and income. For this purpose, panel data for 154 countries worldwide were constructed using annual observations spanning the period from 1995 to 2023.

Carbon emissions are shaped not only by structural factors, such as industrial activity and energy sources, but also by behavioral and distributional dynamics, including inequality and wealth concentration. These factors are often persistent over time, leading to inertia in carbon emission patterns. To account for these persistent and distributionally rooted dynamics, a model is employed that explicitly captures the temporal dependence and nonlinear nature of emissions.

To estimate the dynamic panel data model, the study builds on the methodological advancements introduced by Anderson and Hsiao (1981, 1982), Arellano and Bond (1991), and Blundell and Bond (1998) [52,53,54,55]. Anderson and Hsiao initially proposed using lagged levels of endogenous variables as instruments in first-differenced models. However, their estimator did not exploit all available conditions. Arellano and Bond extended this framework by developing the difference GMM estimator, which incorporates additional orthogonality conditions to improve efficiency. Blundell and Bond later introduced the System GMM estimator, which combines moment conditions from both first-differenced and level equations, thereby improving finite sample properties. To address the potential bias in the two-step covariance matrix, the correction proposed by Bun and Windmeijer (2010) is applied to obtain more reliable standard errors [56].

Given the short time dimension (T = 29 years) and the large cross-sectional dimension (n = 154 countries), traditional estimators such as fixed effects or random effects may yield biased and inconsistent results due to the presence of lagged dependent variables and potentially endogenous regressors. Therefore, the two-step System Generalized Method of Moments (System GMM) estimator is employed. This method produces consistent and efficient estimates in situations where regressors are endogenous and the panel is characterized by a short time dimension and a large number of units.

The choice of System GMM is further justified by the likely endogeneity of several explanatory variables, including income redistribution, which may be subject to reverse causality or omitted variable bias. The estimator addresses these issues by using internal instruments derived from lagged levels and differences. Compared to Difference GMM, the System version improves finite sample properties by incorporating additional moment conditions [55]. In this study, we use the two-step variant of the estimator with the robust option enabled in Stata v17, which applies the Windmeijer correction to standard errors. This ensures valid inference under heteroskedasticity and autocorrelation, and strengthens the reliability of the results [56].

Moreover, the empirical results support the dynamic specification. The lagged dependent variable (${lnCO2}_{i,t-1}$) is statistically significant and consistently less than one in most model specifications, confirming the persistence of emissions and convergence over time. This finding provides empirical justification for the use of a dynamic panel estimator and further validates the methodological choice.

It should be noted that data on carbon emissions of the top 1% are unavailable after 2020. Consequently, the carbon inequality variable—calculated as the ratio of top 1% emissions to national average emissions—is also missing for the same years. As a result, models that use either of these variables as dependent variables are estimated using an unbalanced panel, excluding the years with missing data. This issue of missing data is limited to the “top 1% carbon emissions” and “carbon inequality” variables only. In contrast, the data for average per capita carbon emissions is largely complete throughout the study period and forms the basis for the main estimations. In the System GMM estimations, Stata automatically excludes observations with missing dependent variable values. Therefore, no manual adjustments or imputation methods were applied. This approach was preferred to preserve the integrity of the original dataset and to avoid potential biases that may result from interpolated data.

The System GMM estimator relies on internal instruments—typically lagged levels and differences of endogenous variables—and valid moment conditions to address biases stemming from reverse causality, measurement error, and omitted variable problems. Additionally, the two-step covariance matrix correction is applied following Bun and Windmeijer (2010) to improve inference [56]. All standard errors are clustered at the country level.

We ground the empirical model in the Environmental Kuznets Curve (EKC) hypothesis and estimate it within a dynamic panel framework. It incorporates a lagged dependent variable to capture the persistence of carbon emissions and includes both the level and square of per capita income to test the nonlinear effect of economic growth on emissions. In addition, income redistribution is measured in its logarithmic form to address skewness and allow for elasticity interpretation.

Although country-specific fixed effects are included in all model specifications to control for unobserved heterogeneity across countries, time fixed effects are not explicitly incorporated. This decision was made to avoid instrument proliferation and over-identification issues in the System GMM framework, which may compromise the validity of the Hansen test and the efficiency of the estimates [57]. Moreover, the primary aim of the study is to examine structural drivers of carbon emissions—such as income redistribution and inequality—rather than short-term, time-specific fluctuations. The analysis, therefore, focuses on within-country dynamics and cross-sectional variation while mitigating overfitting concerns associated with excessive time dummies. To assess the robustness of the redistribution effect, the model is estimated separately using both absolute and relative measures of redistribution.

In addition, to explore potential heterogeneity in the redistribution–emissions relationship, we conducted sub-sample estimations by dividing countries into two broad income-based groups: developed countries (high income and upper-middle income) and developing countries (lower-middle income and low income), based on the World Bank’s 2024–2025 income classifications [58]. These estimations allow for a more contextual understanding of the differential impacts of redistribution policies. One country, Venezuela, was excluded from this part of the analysis due to the absence of a formal income classification in the World Bank dataset [58]. As a result, the sub-sample estimations were based on 153 countries instead of the full sample of 154.

The baseline specification is as follows:

$${lnCO2}_{it}=a_0+a_1{lnCO2}_{i,t-1}+{\beta }_1{lnredist}_{it}+{\beta }_2{lngdp}_{it}+{\beta }_3{\left({lngdp}_{it}\right)}^2+{\beta }_4{X}_{it}+e_{it}$$

where

• ${lnCO2}_{it}$ is the natural logarithm of the dependent variable (either per capita carbon emissions or top 1% emissions per capita);
• ${lnCO2}_{i,t-1}$ captures the dynamic (lag) effect of emissions;
• ${lnredist}_{it}$ denotes the logarithm of income redistribution, which is measured using two alternative definitions [49].

  (1)

  Absolute redistribution: the difference between market Gini and net Gini,

  (2)

  Relative redistribution: the share of inequality reduced, calculated as absolute redistribution divided by market Gini;

• ${lngdp}_{it}$ and ${\left({lngdp}_{it}\right)}^2$ are used to test for the EKC hypothesis;
• $X_{it}$ is a vector of control variables (e.g., population, inequality);
• $e_{it}$ is the idiosyncratic error term.

The study examines the impact of income redistribution on carbon emissions using three different dependent variables: (i) per capita carbon emissions; (ii) per capita carbon emissions of the top 1%; and (iii) the ratio of the second to the first, referred to as carbon inequality. Based on the theoretical framework and model specification, the study tests the following hypotheses:

H₁: 

Income redistribution (log) has a negative effect on per capita carbon emissions.

H₂: 

Per capita income (${lngdp}_{it}$) has a positive effect on carbon emissions.

H₃: 

The square of per capita income ${\left({lngdp}_{it}\right)}^2$has a negative effect, consistent with the EKC hypothesis.

4. Findings

This section begins with a descriptive exploration of global CO₂ emissions across income groups, offering context for the subsequent empirical analysis. Before presenting the regression results, it is useful to examine how carbon emissions and emission inequality have evolved over time. Figure 3 and Figure 4 are constructed using data from the World Inequality Database (WID) and Our World in Data (OWID), providing a visual overview of long-term distributional trends in global carbon emissions. Figure 3 presents the historical evolution of global per capita carbon emissions disaggregated by income groups: the top 1% (in red), the top 10% (in green), and the global average (in blue). The black line shows the carbon inequality index, defined as the ratio of top 1% per capita emissions to the overall average.

The figure reveals a persistent and substantial concentration of emissions among the highest income earners. Despite fluctuations in the global average, the per capita emissions of the top 1% remain remarkably high and stable over time. This highlights that the lifestyles of the global elite have remained largely unaffected by broader decarbonization efforts. While the figure underscores the theoretical importance of targeting the top 1% in climate policy, the empirical findings of this study show that income redistribution is not negatively associated with their emissions. Indeed, some models suggest a positive relationship between redistribution and the carbon footprint of the top 1%, indicating that redistribution alone may be insufficient—or, in certain cases, even counterproductive—for mitigating emissions among high-income groups.

While Figure 3 provides a long-term perspective on emission inequality at the global level, Figure 4 focuses specifically on the time frame of the panel dataset (1995–2023), allowing for a closer look at the evolution of emissions among the top 1% relative to the global average. This comparison is particularly important for understanding whether recent decades have seen any meaningful convergence or divergence between elite and average carbon footprints.

Figure 4 plots per capita CO₂ emissions for the top 1% (left axis, red line) alongside the global average (right axis, blue line). The two lines move in parallel until the mid-2000s, after which a clear divergence emerges. Average emissions begin to decline steadily, while emissions by the top 1% plateau at elevated levels.

This pattern offers visual support to the broader finding that carbon emissions remain heavily concentrated among top earners. While average emissions have declined, the persistence of high emissions at the top suggests limited behavioral or structural shifts within this group. These dynamics underscore the need for policy approaches that go beyond income redistribution—particularly when addressing the consumption patterns of the highest-emitting individuals.

We present the empirical estimation results below, building on these visual observations. Before turning to the findings, it is important to briefly review the diagnostic conditions required for the validity of the System GMM estimator [57]. First, the error term must not exhibit second-order autocorrelation. Across all models, the Arellano–Bond test confirms the presence of first-order but not second-order autocorrelation at the 5% level, suggesting consistent and efficient estimation. Second, the number of instruments should remain below the number of observations to avoid instrument proliferation. The Hansen test results—yielding p-values greater than 0.05 but below 1—indicate that the over-identifying restrictions cannot be rejected, confirming the validity of the instruments. Third, the coefficient on the lagged dependent variable should be below one, indicating convergence in the dynamic specification. In the models where the dependent variable is the per capita carbon emissions of the top 1%, the coefficient on the lagged dependent variable is close to or slightly above one. The result suggests a high level of persistence, which is consistent with the structural rigidity of emission patterns among top emitters. While the finding raises some concerns regarding long-run convergence, the results remain interpretable within the context of high-income groups whose consumption behaviors tend to be path-dependent and resistant to short-term change. Taken together, these results indicate that all model specifications satisfy the key diagnostic criteria for System GMM estimation.

With these conditions met,

Table 2. The System GMM, dependent variable: per capita carbon emissions.

	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution			${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$
${lnCO2}_{i,t-1}$	0.964 *** (0.024)	0.966 *** (0.024)	0.963 *** (0.025)	0.970 *** (0.022)	0.972 *** (0.022)	0.970 *** (0.023)
${lnredist}_{it}$	−0.005 ** (0.002)	−0.005 ** (0.002)	−0.008 *** (0.002)	−0.005 ** (0.002)	−0.005 ** (0.002)	−0.009 *** (0.002)
${lngdp}_{it}$	0.206 * (0.107)	0.190 * (0.102)	0.216 * (0.116)	0.181 * (0.099)	0.169 * (0.096)	0.189 * (0.105)
${\left({lngdp}_{it}\right)}^2$	−0.009 ** (0.004)	−0.008 ** (0.004)	−0.010 ** (0.004)	−0.008 ** (0.004)	−0.007 ** (0.003)	−0.009 ** (0.004)
${pop}_{it}$		2.90 × 10−11 *** (9.77 × 10−12)	3.14 × 10−11 *** (1.07 × 10−11)		2.69 × 10−11 *** (9.32 × 10−12)	2.90 × 10−11 *** (9.84 × 10−12)
${gini}_{it}$			−0.001 * (0.000)			−0.001 ** (0.000)
Constant Term	−1.037 * (0.588)	−0.958 * (0.568)	−1.039 * (0.612)	−0.888 (0.549)	0.826 (0.533)	−0.873 (0.555)
Hansen test	17.84 (0.214)	17.89 (0.212)	17.87 (0.213)	6.10 (0.807)	6.24 (0.795)	6.48 (0.774)
AR (1)	−6.97 (0.000)	−6.97 (0.000)	−6.95 (0.000)	−6.97 (0.000)	−6.98 (0.000)	−6.96 (0.000)
AR (2)	0.30 (0.766)	0.30 (0.766)	0.30 (0.765)	0.30 (0.768)	0.30 (0.768)	0.30 (0.767)
Observations	3628	3628	3628	3628	3628	3628

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Cluster standard errors are in parentheses. p-values for AR (1), AR (2), and Hansen tests are in parentheses.

reports the results from dynamic panel models using the natural logarithm of per capita CO₂ emissions as the dependent variable. The main independent variable is income redistribution, measured both in absolute and relative terms. The results indicate that income redistribution has a statistically significant and negative effect on per capita CO₂ emissions across all model specifications. The coefficient ranges between −0.005 and −0.009, indicating that a 1% increase in redistribution is associated with approximately a 0.5% to 0.9% reduction in average per capita CO₂ emissions, ceteris paribus, in the short term. This supports Hypothesis 1, indicating that higher levels of redistribution are associated with lower carbon emissions at the national average level. The magnitude of the effect is slightly stronger when redistribution is measured in relative terms (model 6, −0.009 ***).

As expected under the Environmental Kuznets Curve (EKC) hypothesis, the coefficient on GDP per capita is positive, while its square is negative and significant. This validates Hypotheses 2 and 3, suggesting that CO₂ emissions initially increase with income but decrease beyond a certain income threshold.

Lagged emissions are highly significant and close to unity in all models, pointing to strong inertia in national-level carbon emissions. Population has a small but statistically significant positive effect, likely due to scale. Income inequality (${gini}_{it}$) is marginally significant and negative in some models, though the effect appears relatively weak.

Having examined the effects of redistribution on average per capita carbon emissions, the analysis now turns to its impact on the top 1% of emitters—a group disproportionately responsible for global emissions. This step is essential to assess whether redistribution policies reach and influence high-income, high-emission individuals.

To explore potential heterogeneity in the relationship between income redistribution and per capita carbon emissions, the sample was divided into two income-based groups: developed and developing countries, following the World Bank’s 2024–2025 classification [58]. The results are presented in

Table 3. The System GMM, dependent variable: per capita carbon emissions (sub-samples by income level).

Developed Countries	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution			${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
(n = 93)	(1)	(2)	(3)	(4)	(5)	(6)
Variables	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$
${lnCO2}_{i,t-1}$	0.978 *** (0.018)	0.980 *** (0.018)	0.976 *** (0.020)	0.976 *** (0.019)	0.978 *** (0.019)	0.977 *** (0.018)
${lnredist}_{it}$	−0.007 ** (0.003)	−0.007 ** (0.003)	−0.009 * (0.005)	−0.006 * (0.004)	−0.006 * (0.004)	−0.010 ** (0.005)
${lngdp}_{it}$	0.229 *** (0.075)	0.242 *** (0.070)	0.225 *** (0.068)	0.232 *** (0.082)	0.245 *** (0.077)	0.277 *** (0.068)
${\left({lngdp}_{it}\right)}^2$	−0.011 *** (0.003)	−0.011 *** (0.003)	−0.010 *** (0.003)	−0.011 *** (0.003)	−0.012 *** (0.003)	−0.013 *** (0.003)
		3.20 × 10−11 *** ((1.05 × 10−11))	3.35 × 10−11 *** (1.25 × 10−11)		3.25 × 10−11 *** (1.10 × 10−11)	3.43 × 10−11 *** (1.06 × 10−11)
${gini}_{it}$			−0.000 (0.000)			−0.000 (0.000)
Constant Term	−1.115 *** (0.391)	−1.178 *** (0.367)	−1.091 *** (0.339)	−1.123 *** (0.422)	−1.191 *** (0.398)	−1.332 *** (0.336)
Hansen Test	16.51 (0.283)	17.20 (0.246)	14.31 (0.159)	16.46 (0.286)	17.13 (0.249)	9.31 (0.317)
AR (1)	−4.83 (0.000)	−4.83 (0.000)	−4.86 (0.000)	−4.84 (0.000)	−4.84 (0.000)	−4.84 (0.000)
AR (2)	0.25 (0.801)	0.25 (0.802)	0.25 (0.800)	0.25 (0.801)	0.25 (0.802)	0.25 (0.802)
Observations	2273	2273	2273	2273	2273	2273
Developing Countries	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution			${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
(n = 60)	(1)	(2)	(3)	(4)	(5)	(6)
Variables	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$
${lnCO2}_{i,t-1}$	0.935 *** (0.257)	0.938 *** (0.251)	0.989 *** (0.201)	0.948 *** (0.245)	0.951 *** (0.240)	0.990 *** (0.197)
${lnredist}_{it}$	0.006 (0.045)	0.006 (0.044)	0.001 (0.043)	0.009 (0.052)	0.009 (0.051)	0.001 (0.043)
${lngdp}_{it}$	0.372 (1.37)	0.348 (1.330)	0.128 (1.153)	0.328 (1.330)	0.306 (1.293)	0.124 (1.131)
${\left({lngdp}_{it}\right)}^2$	−0.017 (0.062)	−0.016 (0.060)	−0.007 (0.054)	−0.016 (0.061)	−0.015 (0.059)	−0.007 (0.053)
${pop}_{it}$		2.84 × 10−11 (2.91 × 10−11)	2.24 × 10−11 (2.44 × 10−11)		2.64 × 10−11 (2.86 × 10−11)	2.20 × 10−11 (2.39 × 10−11)
${gini}_{it}$			−0.001 (0.003)			−0.001 (0.002)
Constant Term	−1.867 (7.130)	−1.742 (6.934)	−0.461 (5.757)	−1.608 (6.934)	−1.498 (6.752)	−0.436 (5.726)
Hansen Test	17.58 (0.227)	17.55 (0.228)	17.42 (0.234)	17.66 (0.223)	17.64 (0.224)	17.45 (0.233)
AR (1)	−2.95 (0.003)	−3.00 (0.003)	−3.53 (0.000)	−3.07 (0.002)	−3.12 (0.002)	−3.56 (0.000)
AR (2)	0.14 (0.889)	0.14 (0.889)	0.14 (0.892)	0.14 (0.889)	0.14 (0.889)	0.14 (0.892)
Observations	1328	1328	1328	1328	1328	1328

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Cluster standard errors are in parentheses. p-values for AR(1), AR(2), and Hansen tests are in parentheses. Country groupings follow the World Bank 2024–2025 income classification: high- and upper-middle-income countries are categorized as “developed”, while low- and lower-middle-income countries are categorized as “developing”.

.

In developed countries, both absolute and relative redistribution are consistently associated with a statistically significant reduction in per capita carbon emissions. The coefficient for absolute redistribution ranges from –0.007 to –0.009 and for relative redistribution from –0.006 to –0.010, all significant at the 5% or 1% levels. These findings suggest that redistributive fiscal policies may play an effective role in mitigating emissions in more advanced economies. Additionally, in developed countries, GDP per capita has a positive and significant effect on emissions, while the squared term of GDP per capita is negative and significant—consistent with the Environmental Kuznets Curve (EKC) hypothesis. This indicates a nonlinear growth–emissions relationship. In contrast, the results for developing countries reveal no statistically significant association between income redistribution and per capita carbon emissions. Neither the absolute nor the relative redistribution measures yield meaningful effects. Moreover, both the GDP per capita and the squared term of GDP per capita are statistically insignificant, suggesting the absence of an EKC pattern in these countries. These results highlight important structural differences in the effectiveness of redistribution as a climate policy instrument, depending on countries’ developmental context.

Table 4. The System GMM, dependent variable: per capita carbon emissions of the top 1%.

	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution			${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$
${lnCO2}_{i,t-1}$	1.030 *** (0.033)	1.031 *** (0.033)	1.029 *** (0.030)	1.029 *** (0.032)	1.030 *** (0.032)	1.029 *** (0.030)
${lnredist}_{it}$	0.010 * (0.005)	0.010 * (0.005)	0.007 (0.005)	0.012 ** (0.006)	0.012 ** (0.006)	0.009 (0.006)
${lngdp}_{it}$	0.082 * (0.046)	0.077 * (0.046)	0.091 ** (0.041)	0.085 * (0.046)	0.079 * (0.046)	0.090 ** (0.041)
${\left({lngdp}_{it}\right)}^2$	−0.006 *** (0.002)	−0.006 ** (0.002)	−0.006 *** (0.002)	−0.006 *** (0.002)	−0.006 *** (0.002)	−0.006 *** (0.002)
${pop}_{it}$		2.61 × 10−11 *** (8.93 × 10−12)	2.76 × 10−11 *** (8.04 × 10−12)		2.65 × 10−11 *** (8.83 × 10−12)	2.75 × 10−11 *** (8.13 × 10−12)
${gini}_{it}$			−0.000 (0.000)			−0.000 (0.000)
Constant Term	−0.343 * (0.203)	−0.321 (0.203)	−0.335 * (0.186)	−0.362 * (0.201)	−0.340 * (0.202)	−0.346 * (0.188)
Hansen test	0.25 (0.620)	0.22 (0.641)	0.23 (0.632)	0.24 (0.622)	0.21 (0.644)	0.23 (0.634)
AR (1)	−2.20 (0.028)	−2.20 (0.028)	−2.20 (0.028)	−2.20 (0.028)	−2.20 (0.028)	−2.20 (0.028)
AR (2)	0.80 (0.422)	0.80 (0.423)	0.80 (0.422)	0.80 (0.423)	0.80 (0.423)	0.80 (0.422)
Observations	3227	3227	3227	3227	3227	3227

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Cluster standard errors are in parentheses. p-values for AR (1), AR (2), and Hansen tests are in parentheses.

presents the System GMM estimation results with the per capita carbon emissions of the top 1% as the dependent variable. In contrast to the previous findings for average emissions, the results here indicate a positive and statistically significant relationship between income redistribution and elite emissions in several specifications. In the absolute redistribution models (columns 1 and 2), the coefficient on ${lnredist}_{\mathit{i}\mathit{t}}$ is 0.010 *, while in the relative redistribution models (columns 4 and 5), the effect is slightly stronger at 0.012 **. These coefficients suggest that a 1% increase in redistribution is associated with a 1.0% to 1.2% increase in per capita emissions of the top 1%, ceteris paribus, in the short term. Although not all specifications are statistically significant (e.g., columns 3 and 6), the direction of the effect remains consistently positive. The absence of significance in long-run and robustness models points to potential temporal variation and limits the generalizability of this effect over time.

These findings contradict Hypothesis 1 in the context of the top 1%, indicating that redistribution does not reduce carbon emissions among the highest emitters and may even be associated with increased emissions. One possible interpretation is that redistribution, while benefiting lower-income groups, may unintentionally allow higher-income individuals to maintain or expand carbon-intensive consumption patterns through indirect effects or compensatory behavior.

As in previous models, GDP per capita is positive and statistically significant, while its squared term is negative and significant across all specifications, confirming the Environmental Kuznets Curve (EKC) hypothesis also applies to the top 1%. Population continues to show a small but significant positive association with emissions. The Gini coefficient, however, is not statistically significant in any of the models, suggesting that income inequality may have a limited direct effect on the emissions of the top 1%. Overall, these results emphasize that while redistribution may help reduce average emissions, it does not appear to be an effective tool for limiting emissions among the global elite.

The final part of the analysis focuses on carbon inequality, defined as the ratio of per capita emissions of the top 1% to the national average. This indicator captures the relative gap between elite and average emitters and helps assess whether redistribution plays a role in closing or widening this gap.

Table 5. The System GMM, dependent variable: carbon inequality.

	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution			${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	${\mathit{C}\mathit{O}2\mathit{i}\mathit{n}\mathit{e}\mathit{q}\mathit{u}\mathit{a}\mathit{l}}_{\mathit{i}\mathit{t}}$	${\mathit{C}\mathit{O}2\mathit{i}\mathit{n}\mathit{e}\mathit{q}\mathit{u}\mathit{a}\mathit{l}}_{\mathit{i}\mathit{t}}$	${\mathit{C}\mathit{O}2\mathit{i}\mathit{n}\mathit{e}\mathit{q}\mathit{u}\mathit{a}\mathit{l}}_{\mathit{i}\mathit{t}}$	${\mathit{C}\mathit{O}2\mathit{i}\mathit{n}\mathit{e}\mathit{q}\mathit{u}\mathit{a}\mathit{l}}_{\mathit{i}\mathit{t}}$	${\mathit{C}\mathit{O}2\mathit{i}\mathit{n}\mathit{e}\mathit{q}\mathit{u}\mathit{a}\mathit{l}}_{\mathit{i}\mathit{t}}$	${\mathit{C}\mathit{O}2\mathit{i}\mathit{n}\mathit{e}\mathit{q}\mathit{u}\mathit{a}\mathit{l}}_{\mathit{i}\mathit{t}}$
${CO2inequal}_{i,t-1}$	0.921 *** (0.009)	0.919 *** (0.009)	0.918 *** (0.009)	0.947 *** (0.008)	0.945 *** (0.008)	0.945 *** (0.008)
${lnredist}_{it}$	0.122 (0.142)	0.372 * (0.217)	0.371 * (0.218)	0.107 (0.109)	0.354 ** (0.175)	0.354 ** (0.175)
${lngdp}_{it}$	−11.671 *** (4.084)	−13.166 *** (4.418)	−13.147 *** (4.406)	−6.538 ** (2.662)	−7.612 *** (2.876)	−7.635 *** (2.863)
${\left({lngdp}_{it}\right)}^2$	0.602 *** (0.215)	0.692 *** (0.236)	0.691 *** (0.235)	0.340 ** (0.139)	0.403 *** (0.153)	0.404 *** (0.152)
${pop}_{it}$			−4.68 × 10−10 (5.76 × 10−10)			−1.52 × 10−10 (3.42 × 10−10)
${gini}_{it}$		0.078 ** (0.039)	0.079 ** (0.039)		0.057 * (0.031)	0.058 * (0.031)
Constant Term	56.986 *** (19.327)	59.731 *** (19.931)	59.650 *** (19.887)	31.599 ** (12.748)	33.335 *** (12.752)	33.447 *** (12.702)
Hansen test	18.11 (0.202)	18.05 (0.205)	18.04 (0.205)	18.21 (0.150)	18.20 (0.150)	18.20 (0.150)
AR (1)	−1.16 (0.245)	−1.16 (0.245)	−1.16 (0.245)	−1.16 (0.246)	−1.16 (0.246)	−1.16 (0.246)
AR (2)	−1.02 (0.308)	−1.02 (0.307)	−1.02 (0.308)	−1.02 (0.308)	−1.02 (0.308)	−1.02 (0.308)
Observations	3256	3256	3256	3256	3256	3256

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Cluster standard errors are in parentheses. p-values for AR (1), AR (2), and Hansen tests are in parentheses.

presents the results of System GMM estimations using carbon inequality as the dependent variable. Contrary to theoretical expectations, the results indicate a positive and statistically significant effect of income redistribution on carbon inequality. In the case of absolute redistribution, models (2) and (3) report coefficients of 0.372 * and 0.371 *, respectively. Similarly, relative redistribution also shows a positive and significant relationship (e.g., 0.354 ** in models 5 and 6).

These findings suggest that redistribution may inadvertently increase the carbon gap between the top 1% and the rest of the population rather than reducing it. This result raises critical questions about the equity and efficiency of current redistributive mechanisms in addressing high-emission lifestyles.

Other explanatory variables yield additional insights. GDP per capita has a negative and significant effect, while its square term is positive and significant—pointing to a nonlinear relationship consistent with a reversed Environmental Kuznets Curve (EKC) dynamic: at lower income levels, carbon inequality narrows, but it begins to widen again beyond a certain income threshold.

Income inequality (${gini}_{it}$) is positively associated with carbon inequality, indicating that broader economic disparities are mirrored in environmental inequalities. The population variable is not statistically significant in any specification.

To assess long-term relationships,

Table 6. The System GMM, The Long-Run Coefficients.

Per Capita Carbon Emission	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution			${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$
${lnCO2}_{i,t-1}$	27.111 (19.431)	29.129 (21.866)	26.588 (19.694)	32.893 (25.921)	34.974 (28.730)	32.598 (26.064)
${lnredist}_{it}$	−0.159 (0.157)	−0.170 (0.173)	−0.232 (0.173)	−0.178 (0.206)	−0.188 (0.225)	−0.321 (0.269)
${lngdp}_{it}$	5.812 *** (1.209)	5.753 *** (1.274)	5.981 *** (1.232)	6.163 *** (1.499)	6.086 *** (1.555)	6.377 *** (1.580)
${\left({lngdp}_{it}\right)}^2$	−0.267 *** (0.071)	−0.264 *** (0.076)	−0.279 *** (0.073)	−0.290 *** (0.091)	−0.287 *** (0.095)	−0.304 *** (0.097)
${pop}_{it}$		8.73 × 10−10 *** (2.94 × 10−10)	8.66 × 10−10 *** (2.95 × 10−10)		9.67 × 10−10 *** (3.35 × 10−10)	9.73 × 10−10 *** (3.31 × 10−10)
${gini}_{it}$			−0.024 *** (0.008)			−0.033 ** (0.016)
Constant Term	−29.175 *** (4.645)	−28.885 *** (4.863)	−28.690 *** (4.462)	−30.130 *** (5.319)	−29.737 *** (5.456)	−29.338 *** (4.938)
Per Capita Carbon Emissions of the Top 1%	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution			${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
	(1)	(2)	(3)	(4)	(5)	(6)
Variables	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$	${\mathit{l}\mathit{n}\mathit{C}\mathit{O}2}_{\mathit{i}\mathit{t}}$
${lnCO2}_{i,t-1}$	−33.761 (35.658)	−32.505 (32.864)	−35.246 (36.272)	−34.668 (37.112)	−33.347 (34.151)	−35.146 (36.099)
${lnredist}_{it}$	−0.348 (0.364)	−0.345 (0.347)	−0.267 (0.350)	−0.414 (0.412)	−0.410 (0.392)	−0.312 (0.407)
${lngdp}_{it}$	−2.712 (3.896)	−2.441 (3.470)	−3.123 (3.925)	−2.865 (4.096)	−2.580 (3.641)	−3.076 (3.862)
${\left({lngdp}_{it}\right)}^2$	0.208 (0.229)	0.192 (0.205)	0.234 (0.232)	0.218 (0.242)	0.202 (0.215)	0.231 (0.229)
${pop}_{it}$		–8.22 × 10−10 *** (2.81 × 10−10)	–9.45 × 10−10 *** (2.75 × 10−10)		–8.58 × 10−10 *** (2.86 × 10−10)	–9.39 × 10−10 *** (2.78 × 10−10)
${gini}_{it}$			0.030 * (0.016)			0.024 * (0.014)
Constant Term	11.239 (16.443)	10.121 (14.671)	11.502 (16.007)	12.190 (17.562)	11.012 (15.658)	11.844 (16.357)

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Cluster standard errors are in parentheses.

reports the long-run coefficients derived from the dynamic models. The effect of redistribution on both average per capita emissions and top 1% emissions is negative in direction but statistically insignificant across all specifications. These results suggest that the emission-reducing potential of redistribution policies may be limited over the long term. The EKC relationship, however, remains robust: GDP per capita has a positive and significant effect, while its squared term is negative and significant, consistent with the hypothesis.

Table 7. The System GMM, Robustness Check 1.

	1960–2023				1960–2023 (Long-Run Coefficients)
	Per Capita Carbon Emissions		Per Capita Carbon Emissions of the Top 1%		Per Capita Carbon Emissions		Per Capita Carbon Emissions of the Top 1%
Variables	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
${lnCO2}_{i,t-1}$	0.961 *** (0.031)	0.960 *** (0.031)	1.023 *** (0.024)	1.023 *** (0.024)	25.032 (21.184)	24.618 (20.646)	−42.753 (42.472)	−42.828 (42.682)
${lnredist}_{it}$	−0.009 *** (0.003)	−0.009 ** (0.004)	0.005 (0.005)	0.006 (0.005)	−0.247 (0.235)	−0.254 (0.251)	−0.220 (0.269)	−0.255 (0.311)
${lngdp}_{it}$	0.256 (0.160)	0.261 (0.161)	0.065 * (0.037)	0.064 * (0.037)	6.684 *** (1.402)	6.687 *** (1.397)	−2.743 (3.638)	−2.715 (3.623)
${\left({lngdp}_{it}\right)}^2$	−0.012 * (0.007)	−0.012 * (0.007)	−0.004 ** (0.002)	−0.004 ** (0.002)	−0.316 *** (0.082)	−0.316 *** (0.082)	0.204 (0.207)	0.203 (0.206)
${pop}_{it}$	3.94 × 10−11 ** (1.60 × 10−11)	4.01 × 10−11 ** (1.61 × 10−11)	3.74 × 10−11 *** (9.92 × 10−12)	3.72 × 10−1 *** (1.00 × 10−11)	1.03 × 10−9 ** (4.16 × 10−10)	1.03 × 10−9 ** (4.12 × 10−10)	−1.56 × 10−9 *** (4.14 × 10−10)	−1.56 × 10−9 *** (4.19 × 10−10)
${gini}_{it}$	−0.001 (0.000)	−0.001 (0.000)	−0.000 (0.000)	−0.000 (0.000)	−0.029 *** (0.010)	−0.033 ** (0.013)	0.035 (0.023)	0.030 (0.021)
Constant Term	−1.215 (0.817)	−1.220 (0.824)	−0.250 (0.167)	−0.257 (0.167)	−31.654 *** (5.183)	−31.274 *** (4.843)	10.443 (15.239)	10.764 (15.553)
Hansen test	0.05 (0.817)	0.07 (0.799)	0.42 (0.517)	0.42 (0.517)
AR (1)	−2.30 (0.021)	−2.30 (0.021)	−2.78 (0.005)	−2.78 (0.005)
AR (2)	1.10 (0.272)	1.10 (0.272)	1.23 (0.219)	1.23 (0.219)
Observations	5458	5458	4356	4356	5458	5458	4356	4356

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Cluster standard errors are in parentheses. p-values for AR (1), AR (2), and Hansen tests are in parentheses.

presents robustness checks using an extended sample period from 1960 to 2023. The short-run results are largely consistent with the main findings: income redistribution significantly reduces average per capita emissions, while its effect on the top 1% remains statistically insignificant. In the long-run models, the effect of redistribution loses statistical significance, both for the average and for the top 1%. However, the EKC hypothesis is robust across specifications.

Table 8. The System GMM, Robustness Check 2.

	2000–2023				2010–2023
	Per Capita Carbon Emissions		Per Capita Carbon Emissions of the Top 1%		Per Capita Carbon Emissions		Per Capita Carbon Emissions of the Top 1%
Variables	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Absolute Redistribution	${\mathit{l}\mathit{n}\mathit{r}\mathit{e}\mathit{d}\mathit{i}\mathit{s}\mathit{t}}_{\mathit{i}\mathit{t}}$ = Relative Redistribution
${lnCO2}_{i,t-1}$	0.961 *** (0.028)	0.951 *** (0.026)	0.993 *** (0.031)	0.985 *** (0.039)	0.849 *** (0.031)	0.872 *** (0.033)	0.728 *** (0.065)	0.728 *** (0.068)
${lnredist}_{it}$	−0.007 ** (0.003)	−0.009 * (0.006)	0.002 (0.004)	0.003 (0.004)	−0.015 * (0.008)	−0.020 * (0.012)	0.008 (0.011)	0.018 (0.011)
${lngdp}_{it}$	0.198 * (0.117)	0.233 * (0.124)	0.095 *** (0.031)	0.105 *** (0.030)	0.505 *** (0.131)	0.425 *** (0.157)	−0.181 (0.144)	−0.116 (0.121)
${\left({lngdp}_{it}\right)}^2$	−0.007 ** (0.003)	−0.010 * (0.005)	−0.005 ** (0.002)	−0.005 * (0.002)	−0.019 *** (0.005)	−0.016 ** (0.007)	0.022 ** (0.009)	0.018 ** (0.008)
${pop}_{it}$	3.32 × 10−1 ** (1.44 × 10−11)	3.82 × 10−11 ** (1.63 × 10−11)	3.85 × 10−11 *** (1.33 × 10−11)	3.62 × 10−11 *** (1.21 × 10−11)	7.72 × 10−11 *** (2.44 × 10−11)	8.17 × 10−11 *** (2.79 × 10−11)	1.17 × 10−10 *** (4.00 × 10−11)	1.09 × 10−10 *** (4.07 × 10−11)
${gini}_{it}$	−0.000 (0.000)	−0.001 (0.000)	−0.000 (0.000)	6.93e−06 (0.000)	−0.002 * (0.001)	−0.002 (0.001)	0.007 *** (0.002)	0.007 *** (0.002)
Constant Term	−0.963 (0.633)	−1.140 * (0.641)	−0.378 *** (0.130)	−0.444 *** (0.120)	−2.769 *** (0.684)	−2.290 *** (0.795)	0.410 (0.592)	0.117 (0.503)
Hansen test	4.19 (0.381)	25.98 (0.166)	17.91 (0.161)	0.71 (0.399)	3.17 (0.530)	25.24 (0.192)	22.76 (0.064)	15.34 (0.082)
AR (1)	−6.87 (0.000)	−6.93 (0.000)	−5.44 (0.000)	5.32 (0.000)	−4.98 (0.000)	−5.01 (0.000)	−4.40 (0.000)	−4.43 (0.000)
AR (2)	−0.48 (0.633)	−0.48 (0.634)	−0.08 (0.936)	−0.07 (0.942)	−1.07 (0.283)	−1.07 (0.283)	−1.75 (0.080)	−1.77 (0.077)
Observations	2996	2996	2622	2622	1585	1585	1277	1277

Notes: ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. Cluster standard errors are in parentheses.

presents additional robustness checks by estimating the models over two overlapping time frames: the full post-2000 period (2000–2023) and the more recent post-2010 period (2010–2023). The results for average per capita emissions show that the negative effect of redistribution persists in both windows and becomes stronger and statistically significant in the 2010–2023 period—particularly in both absolute (−0.015 *) and relative (−0.020 *) redistribution models. This finding suggests an increased effectiveness of redistributive policies in curbing average emissions during the last decade.

In contrast, the results for the top 1% emissions remain statistically insignificant in both periods, indicating that redistribution continues to have limited influence on high-income, high-emission individuals. These findings imply a growing divergence between the redistributive effects across income groups and underscore the structural rigidity of elite carbon consumption.

Overall, the results confirm that income redistribution is effective in reducing average per capita CO₂ emissions, supporting Hypothesis 1. However, redistribution appears to be ineffective—or even counterproductive—when it comes to curbing emissions among the top 1% and may contribute to widening carbon inequality. In line with Hypotheses 2 and 3, the analysis also reveals a nonlinear relationship between income and emissions, providing support for the Environmental Kuznets Curve (EKC) hypothesis.

5. Discussion

The environmental problem has various ways of impacting life. One aspect of the environmental issues is income redistribution and its link to carbon dioxide emissions. The issue leads to a debate in academia regarding whether the distribution of income affects carbon emissions and whether there is a connection between people’s income levels and carbon emissions.

From this perspective, the study has three dimensions for analyzing the results. The overall per capita carbon emissions at the national level, the per capita carbon emissions of the top 1% income group across countries, and the ratio of the top 1%’s per capita carbon emissions to the national average are referred to as carbon inequality between 1995 and 2023. The panel data for 154 countries represent the sample size. In the literature, the income inequality, distribution, and the effect on carbon emissions is a comprehensive and debated issue.

One result from our analysis reveals a nonlinear relationship between income and emissions, providing support for the EKC hypothesis. This result promotes the theoretical literature of the inverted U-shaped relationship with per capita income and environmental deterioration [14,15,16].

In the literature on income inequality and carbon emissions, there are studies with divergent results. While some of the research studies have indicated a positive relationship between income inequality and carbon emissions [33,34,35,59], other research studies have demonstrated an inverse relationship [23,29,32]. As evaluated, the results vary about this relationship, and arguments are still continuing.

Another discussion was made about redistribution. Our results point out that an increase of 1% in redistribution is associated with a 0.5% to 0.9% reduction in average per capita CO₂ emissions in the short term. The results align with various studies [40,42,44].

Our study develops both the time range (1995–2023) using actual data and a wide dataset that includes 154 countries. As the carbon emissions of the countries change over time, years-long analysis will contribute to the existing literature. In addition to the literature focusing more on carbon taxes and redistribution on a country basis [43,60,61], this study concentrated on the redistribution effect of carbon emissions at a multi-country level.

Our study also highlights the differences in carbon emissions across various income groups. The results prove that individuals do not contribute to emissions in the same way. Our analysis covers multiple years instead of just one year [26,62] and includes multiple countries rather than focusing on a single-country analysis on low-, middle-, and high-income levels [25]. This broader approach allows for a more comprehensive understanding of the complex dynamics governing emissions and their redistribution, ultimately providing insights that can inform policy decisions aimed at reducing carbon footprints across different socio-economic contexts. By examining the interplay between income levels and carbon emissions, we can better address the equity implications of climate change initiatives.

This finding reinforces the idea that redistributive fiscal policies can play a role in climate mitigation by lowering average emissions, particularly when designed within supportive institutional and economic contexts. However, the effectiveness of redistribution may not be uniform across all country settings. Our sub-sample analysis revealed that this negative relationship between redistribution and per capita CO₂ emissions is statistically significant only in developed countries. In contrast, in developing countries, the estimated effects were statistically insignificant. This suggests that the environmental impact of redistribution depends not only on the magnitude of policy intervention but also on underlying structural capacities, such as administrative quality and policy implementation. The sub-sample analysis reveals a key policy asymmetry: income redistribution has a statistically significant and negative impact on emissions in developed countries, while the effect is weaker and statistically insignificant in developing countries. This suggests that redistribution policies may be more effective in reducing emissions where fiscal institutions are stronger and administrative capacities are higher. In contrast, in developing economies, redistribution alone may not suffice unless accompanied by structural reforms, targeted investments, and capacity-building in environmental governance. Therefore, policy frameworks should be context-sensitive: in advanced economies, redistribution can be a viable environmental tool, while in less developed settings, it should be integrated with broader developmental and institutional strategies.

In several model specifications, the long-run coefficients derived from the dynamic estimations are statistically insignificant. This outcome may stem from the structural volatility and institutional heterogeneity across countries, which weaken the persistence of redistributive effects over time. While redistribution policies may lead to short-term reductions in emissions through immediate behavioral or fiscal responses, sustaining these effects likely depends on long-term regulatory stability and policy coherence. Furthermore, limited within-country temporal variation in redistribution indicators may reduce the statistical power to detect significant long-run impacts. Therefore, the insignificance of long-run effects should not be interpreted as a lack of causal relevance but rather as a reflection of complex, uneven, and possibly time-variant relationships between redistribution and emissions.

The income level of the people is another topic to discuss about carbon emissions. The income level and its environmental effect and if there is a connection between the income level and carbon emissions are relatively new in the literature. The more studies are conducted on this topic, the more progress will be made in clarifying the issue. Our finding suggests that a 1% increase in redistribution is associated with a 1.0% to 1.2% increase in per capita emissions among the top 1% in the short term. Moreover, the lowest-income group is affected by environmental issues [24]; the lifestyle of the wealthy is more carbon-intensive [25]. Other studies conclude the positive relationship between income and carbon emissions [33,36]. Our study confirms the findings of Chancel et al. (2023) and suggests that the wealthiest 1% are primarily responsible for global carbon emissions [26]. Undoubtedly, a decrease in the income of wealthy people would lead to a decline in their consumption. The public policy is effective if it raises the living standards of low-income earners without harming the environment.

The most distinctive and strongest aspects of this study are the impact of income redistribution policies on carbon emissions and the identification that the wealthiest 1% of the population in countries significantly contributes to carbon emissions relative to others.

6. Study Limitations and Directions for Future Research

This study, despite its empirical rigor and extensive country coverage, acknowledges several limitations that offer fruitful directions for future research.

Another limitation of the study relates to the exclusion of control variables, such as energy structure or industrial composition. Although these variables may influence carbon emissions, reliable and comparable data covering a wide set of 154 countries over nearly three decades are highly limited—especially for datasets that can be aligned with the multiple carbon-related indicators used in this study. Moreover, the inclusion of multiple sectoral controls in System GMM models can increase the risk of instrument proliferation and reduce the reliability of diagnostic tests. Given these constraints, the model focuses on core socio-economic variables that are widely available, consistently reported, and commonly used in the literature.

First, the analysis draws on secondary sources like WID, OWID, and SWIID, which differ in methodology and data construction. Combining these sources—especially for composite indicators like “carbon inequality”—may mask inconsistencies and introduce bias. Future research should prioritize more consistent, micro-level data to better assess carbon inequality.

Second, although System GMM is well suited for dynamic panel estimation, the exclusion of time fixed effects to avoid instrument proliferation may reduce the model’s ability to account for global shocks or time-specific policy changes. Alternative estimation strategies could help address this limitation.

Third, the study conceptualizes income redistribution through two measures—absolute and relative—but does not distinguish between the underlying policy instruments, such as taxes and transfers. Disentangling these instruments in future analyses could help identify which types of redistributive policies are more conducive to reducing emissions and closing the carbon gap. Moreover, mixed-method or institutional approaches may shed light on the behavioral and structural mechanisms through which redistribution influences environmental outcomes.

Fourth, while this study includes a wide sample of 154 countries, the pooled estimation approach treats all countries as a single analytical unit without accounting for variation across different economic, fiscal, or institutional contexts. To partially address this concern, we conducted sub-sample analyses by dividing countries into two income groups—developed and developing—based on the World Bank’s 2024–2025 country classifications by income level [58]. These sub-group estimations provide initial insights into how the effectiveness of redistribution may vary across different developmental contexts. One country (Venezuela) was excluded from the sub-sample analysis due to the lack of a formal income classification. Therefore, the sub-group results are based on 153 countries instead of the full sample. Future research could expand this approach by stratifying countries more granularly—such as into low-, middle-, and high-income groups—or by clustering them based on fiscal capacity or climate vulnerability to further investigate the contextual heterogeneity in redistribution–emissions dynamics.

However, it should be noted that the sub-sample estimations could only be conducted for average per capita emissions. Attempts to apply the same approach to top 1% emissions and carbon inequality variables resulted in model instability, including convergence failures, inconsistent coefficient signs, and weak diagnostic statistics. As a result, these estimations were limited to the full-sample level only.

Finally, this study focuses on average and top 1% emissions, overlooking the full distribution of carbon responsibility. Future research could incorporate decile-level and longitudinal data to examine how income and emissions interact across population segments over time.

7. Conclusions

This study examined the relationship between income redistribution and carbon emissions across 154 countries from 1995 to 2023, using a dynamic panel approach with System GMM estimations. The findings indicate that income redistribution—measured in both absolute and relative terms—is associated with reductions in average per capita carbon emissions. This suggests that redistributive policies may have the potential to support national-level climate mitigation efforts.

However, the study also finds that redistribution does not appear to significantly reduce emissions among the top 1% income group. In some cases, a positive association was observed, and redistribution was found to be linked with rising carbon inequality. These results, while not conclusive in establishing causality, point to the challenges of addressing emissions concentrated among high-income groups, where consumption tends to be more persistent and carbon-intensive.

Based on these findings, one key policy implication emerges: while redistribution may help reduce average emissions, more targeted tools may be needed to influence the emissions of top emitters. Progressive carbon pricing or levies on high-emission luxury consumption—such as private aviation, large vehicles, or energy-intensive real estate—could be considered as complementary policy instruments.

Beyond its empirical results, this study contributes to the literature by linking redistribution dynamics with both average and elite-level carbon emissions in a global panel setting. It also highlights the importance of considering distributional dimensions—such as carbon inequality—when evaluating climate policies. Future research could extend this work by incorporating micro-level consumption data, differentiating between specific fiscal instruments (e.g., taxes vs. transfers) or exploring the institutional pathways through which redistribution may affect environmental outcomes over time.

Additionally, the sub-sample analyses underscore that the effectiveness of redistribution in reducing carbon emissions varies across different economic contexts. The relationship was statistically significant only in developed countries, suggesting that redistributive environmental policies may depend heavily on the underlying institutional and structural capacities of countries.

In sum, while income redistribution can contribute to environmental goals, it may be most effective when combined with strategies that directly address emissions at the top of the distribution. A balanced approach linking equity and climate objectives may offer a more comprehensive path toward sustainable development.