Investigating the environmental Kuznets curve modified with HDI: evidence from a panel of eco-innovative countries

Abstract

Considering the increase in polluting emissions, the importance of determining the relationship between economic development and environmental quality is a crucial issue in achieving sustainable development goals. This study explores and empirically tests the validity of the Environmental Kuznets Curve (EKC) hypothesis in a panel data of 39 countries is assembled over the period 1993–2018. For this purpose, the Human Development Index is used as a measure of development, contrary to classical studies in which GDP growth was determined as an indicator of economic growth. Also using by two different measures of pollutant emissions, such as methane (${{\text{CH}}}_{4}$) and carbon (${{\text{CO}}}_{2}$) emissions, examining to the validity of the EKC hypothesis. The analysis method consists of the fixed effect and robustness estimation of the panel data model, and employed the U-test approach in the study. Empirical findings in this study show that strong evidence to support an inverted U-shaped relationship between CO₂ emissions and human development, although the U-test results point at mixed results for CH₄ emissions. In addition, it is detected that negative relationship between environmental related technological innovation and pollutants emission such as ${{\text{CH}}}_{4}$ and ${{\text{CO}}}_{2}$, which highlights the importance of promoting innovative technologies in struggle global warming.

1 Introduction

Because of the unavoidable increase in emissions in recent decades, environmental issue such as global warming and climate change have reached a levels that cannot be ignored. According to report of the Intergovernmental Panel on Climate Change (IPCC, 2023), atmospheric carbon dioxide ${({\text{CO}}}_{2})$ concentrations reached the highest levels seen in at least two million years. Also, methane concentrations exceeded the highest values seen in 800,000 years. The Environmental Protection Agency (2023) emphasizes the significant role of greenhouse gas (GHG) emissions in global warming. Accordingly, ${{\text{CO}}}_{2}$ comprises 76% of global greenhouse gas emissions, while methane (${{\text{CH}}}_{4}$) is comprised of 16%. The substantial relationship between high greenhouse gas emissions and climate change directly affects public health, leading to a global decline in the quality of the environment and impacting that which is defined as healthy living (Dritsaki & Dritsaki, 2023). Additionally, climate change affects biodiversity due to changes in optimum temperature levels and compromises the survival chances of many species (Abbass et al., 2022). To mitigate the issue of high greenhouse gas emissions, numerous nations collectively entered into the Paris Agreement in 2015. However, many member states still face environmental issues such as the depletion of natural resources, desertification, and poor urban air quality (Zhang et al., 2023; Khan et al., 2022; Dong et al., 2018). At the stage of shaping a sustainable economy model, uncovering the negative effects of these problems becomes the focus of attention of researchers. In this context, the EKC hypothesis, which is widely discussed in the literature, assumes that there is a relationship between environmental pollution and economic growth. According to the EKC hypothesis, environmental pollution increases up to a point in the first stage with economic growth, but when the turning point is reached environmental pollution tends to decrease, thus suggesting that there is an inverted U-shaped relationship between environmental pollution and economic growth (Panayotou, 1993). Therefore, determining the factors that cause the inverted U-shaped relationship becomes an important study subject in terms of sustainable development.

There are many indicators such as carbon dioxide (${{\text{CO}}}_{2}$), nitrogen dioxide (${{\text{NO}}}_{2}$), particulate matter (PM), methane (${{\text{CH}}}_{4}$) and sulfur dioxide (${{\text{SO}}}_{2}$), which considered as the environmental pollution indicators in empirical studies. In most recent studies to examining the nexus between economic growth and environmental pollution, researchers have focused on the release of ${{\text{CO}}}_{2}$ emissions as a indicator of environmental degradation (Salahuddin et al., 2015; Wang et al., 2016). On the other hand, ${{\text{CO}}}_{2}$ emissions have been the most widely used air pollution indicator in previous studies focusing on EKC, as it is constitutes the majority of total greenhouse gas (GHG) emissions. However CO₂ emission is only one of the GHG emissions such as ${{\text{CH}}}_{4}$ and ${{\text{NO}}}_{2}$, which is an indicator of environmental degradation. ${{\text{CH}}}_{4}$ emissions, one of these pollutants, are less persistent in the atmosphere than others (GHG), but it is ten times stronger than CO₂ in heating the atmosphere. Although the transition to industrial production with the development of the economy causes an increase in the amount of CO₂ emissions, it should not be overlooked that the analyzes, carried out without considering other GHG emissions, will be restrictive (Cho et al., 2014; Yusuf et al., 2020). According to the UNEP (2021) report, ${{\text{CH}}}_{4}$ is the most important second global GHG and it seems to more harmful than ${{\text{CO}}}_{2}$ in acting as a cause of global warming over the next years. In addition, researches of the Intergovernmental Panel on Climate Change (IPCC) emphasize that ${{\text{CH}}}_{4}$ emissions have a share of about a quarter in global warming and that human-induced emissions, which cause more than half of this rate, have a great role in preventing climate change. The graphs in Figs. 2 and 3 (see Appendix 1) show that countries with high ${{\text{CO}}}_{2}$ emissions also have high ${{\text{CH}}}_{4}$ emissions. Considering this perspective, it raises the question that the level of economic and human capital required to reduce the amount of ${{\text{CO}}}_{2}$ emissions may actually be decisive for reducing the amount of ${{\text{CH}}}_{4}$, which is the second largest GHG.

Most of the remarkable studies examining the EKC hypothesis focus heavily on the narrow relationship between environmental pollution and economic growth. On the other hand, the factors associated with environmental pollution are broader. While energy consumption is one of the basic inputs of production processes, it is directly related to environmental degration in terms of emissions resulting from economic activities. According to the studies by Kaika and Zervas (2011) and Kasman and Duman (2015), considering at energy consumption as an indicator of environmental pollution is plausible, as ${{\text{CO}}}_{2}$ emissions are largely directly related to energy consumption. To achieve higher economic growth, industrial processes require energy inputs such as fossil fuels and natural resources that release harmful pollutants directly into the natural environment (Al-Mulali, 2014; Bashir et al., 2021). Studies such as Ike et al. (2020), Altıntaş and Kassouri (2020) and Gill et al. (2018) indicate that there is an evidence denoting that the major reason for the increase in CO₂ emissions can be relevant to energy consumption, especially using of the fossil fuels. Thus, the need for energy supply becomes an important topic in terms of the sustainability of the current economic and social functioning.

In addition the factors such as energy consumption, population and environmental-related technological improvement activities, human capital also plays a decisive role for environmental pollution (Shahbaz et al., 2013; Gormus & Aydin, 2020, Ahmed et al., 2022). The importance of the this relationship between human capital and environmental degradation can be seen in the studies conducted by (Zhang et al., 2021, Farhani et al., 2014). Human capital might contribute to the understanding of environmental degration issues (İçen & Çil, 2023). The reason for this is that human development indicators can have an impact on environmental quality in a precise both of directly and indirectly. One of the most essential aspects of economic success and development is the development of human capital (Hanif et al., 2020; Salim et al., 2017). According to Dinda (2004), with an increase in income, people achieve higher living standards and are more concerned about the quality of the environment they are a part of, and as a result of the demand for a better environment leads to structural changes in the economy that reduce environmental degradation. In this context, people’s income, education level and quality of life will directly affect their demands for a cleaner environment and will enable the adoption of cleaner technologies. Another factor that will enable clean technologies to be adapted to economic and social life is the adoption of environmentally friendly technological improvements. Additionally, environmentally focused technological innovation is considered an effective mechanism for reducing emissions and promoting sustainable growth (Amin et al., 2023). A wealthy nation can achieve more technological progress with economic growth, as the replacement of outdated and polluting technologies with upgraded new and cleaner technologies that improve environmental quality is associated with R&D expenditures (Adedoyin et al., 2020; Awan & Azam, 2021; Balsalobre et al., 2015; Bilgili et al., 2021). Consequently it is expected that people will be more sensitive to environmental degradation and safety in countries with developed economic structure and high human capital.

The environmental aspect of nexus between ${{\text{CO}}}_{2}$ emissions and economic growth has been extensively investigated through the EKC hypothesis (Danish et al., 2019; Cheik et al., 2021; Churcill et al., 2020). Contrary to other GHG emissions, there are relatively few studies examining the relationship between ${{\text{CH}}}_{4}$ emissions and economic growth with the EKC hypothesis in the literature. The inspirition of this paper that the EKC literature generally considers the traditional measure of environmental pollution of ${{\text{CO}}}_{2}$ emissions but regrettably there is insufficient evidence under the EKC hypothesis for ${{\text{CH}}}_{4}$ emissions which has the second largest impact on climate change. Recent studies (Aydin et al., 2023, Wang et al., 2023) have sought to determine the effect of human development on the shape of the EKC, focusing on how these factor might flatten or increase the scope of the curve. However, while most studies focus on economic growth, there is not enough research that incorporates the square of the human development variable in the equation as a determinant of the Environmental Kuznets Curve (EKC) form. Although there are studies on income in the existing literature on environmental pollution, considering that the regulatory role of human development is not adequately addressed, this study can be seen as an answer to this deficiency. Moreover, this paper investigates the turning point using the human development index within the context of sustainable environmental quality. Our basis contribution to the literature lies in assessing the influence of the human development index on the validity of the Environmental Kuznets Curve (EKC) and examining in detail the impact of various variables, including technological innovations, on ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions across a broad panel of countries. Thus, focusing on all the countries that can be reached within the framework of the analysis presents the importance of determining an environment-oriented perspective within the scope of sustainable economic development. Hence our empirical analysis on the modified with human capital environmental Kuznets curve is performed on a newest data covering 39 countries for the period 1993–2018.

In this paper, we hypothesize that the relationship between human development and emission levels is nonlinear, and at higher levels of human development, this may begin to contribute substantially to a cleaner environment. Both models, with pollution indicators such as CO₂ emissions and ${{\text{CH}}}_{4}$ emissions, provide significant evidence for the existence of the human development-based EKC. In methodological terms, the study employs panel causality test, Lind and Mehlum (2010) U-shape relationship test and panel-corrected standard errors (PCSE) and The Driscoll–Kraay (DK) methods.

The remaining sections are as follows. Section 2 briefly describes the findings from empirical studies in the literature. Section 3 examines the models, estimation methods, and data sources used to test the EKC hypothesis. Empirical results obtained in Sect. 4 and related discussions are presented. The last section includes the conclusion comments.

2 Literature review

The applicability of the EKC hypothesis is based on the mechanism presented by Kuznets (1955) in his seminal article, which allows for the examination of the inverted U-shaped relationship between economic growth and income inequality. Afterward, Grossman and Krueger (1991) examined the specific linkage between economic growth and environmental pollution within the framework of the theoretical mechanism proposed by Kuznets (1955). The studies of Grossman and Krueger (1991) provided the opportunity to consider the relationship between environment and economic growth through a different perspective and pioneered further studies (Leal and Marquez, 2022). In the early studies Panayotou (1993), Selden and Song (1994), Holtz-Eakin and Selden (1995) and Agras and Chapman (1999) the focus was on the impact of per capita income on per capita emissions through various methods. While the pollution itself is basically a simple idea, since the scope of the indicators considered as pollution indicators is wide, an alternative way to annihilate the complexity in a single pollutant selection point is to create a more comprehensive measure of pollution through different pollutants (Brajer et al., 2011). Thus, the context of the EKC hypothesis has been improved, and various studies (Shafik and Bandyopadhyay (1992), Suri and Chapman (1998), List and Gallet (1999)) have incorporated different dependent and independent variables into the EKC model. Some of the applications (Fodha & Zaghdoud, 2010, Khan et al., 2016), investigating the validity of the EKC hypothesis have focused on pollutants such as ${{\text{SO}}}_{2}$ or $PM$ emissions, while many studies (Acaravci & Ozturk, 2010; Apergis & Payne, 2009), address $GHG$ emissions and examine the impact of factors like energy consumption and population growth. Recent studies in the literature, Rauf et al. (2018), Li et al. (2016), Abid (2016), Barra and Zotti (2018), Gyamfi et al. (2021), Sarkodie and Strezov (2018) examines the validity of the EKC hypothesis and thus the relationship between per capita income and emission. The findings highlight that, like income growth, the increase in energy demand caused by population growth also has an impact on emissions. In addition to these Wang (2012) for 98 countries; Hanif and Gago-de-Santos (2017) for 86 developing countries; Omri et al. (2015) for MENA countries; Pao and Tsai (2010) for BRIC countries; Apergis and Ozturk (2015) for 14 ASEAN countries; Apergis (2016) for 15 OECD countries; Rodríguez et al. (2016) for 15 OECD countries; Liu et al. (2017) for ASEAN-4 contries; Bibi and Jamil (2021) for 6 region; Xia et al. (2022) for 67 countries, Ahmad et al. (2021) for 11 developing countries, Apergis and Payne (2010) for 11 countries; Du and Xie (2020) for 145 countries, Angulo et al. (2018) for 182 countries; Koilo (2019) 11 Eastern European and Central Asian countries; Mania (2020) for 98 countries; Dogan and Inglesi-Lotz (2020) for 7 European countries examines the relationship between income and ${{\text{CO}}}_{2}$ emissions within the framework of the EKC hypothesis and verify the existence of an inverted U-shaped relationship between income and ${{\text{CO}}}_{2}$ emissions in various countries and regions. On the other hand Aslanidis and Iranzo (2009) 77 nonOECD countries; Adu and Denkyirah (2017) for selected West African countries; Arouri et al., (2012) for MENA countries; Ozcan (2013) for Middle East; Bölük and Mert (2014) for 16 European Union countries; Özokcu and Özdemir (2017) for 26 OECD countries; Abid (2017) for 58 countries; Hove and Tursoy (2019) for 24 emerging economies; Ng et al. (2020) for 76 countries; Saidi and Mbarek (2017) for 19 emerging economies; Altıntaş ve Kassouri (2020) for 14 European countries; Beyene and Kotosz (2020) for 12 East African countries found controversial evidence not supporting the validity of EKC hypothesis. These studies also differ due to different model selection, data period considered, explanatory variables and their number or form of EKC. Therefore, some studies reach the conclusion that the EKC is valid, while others do no sign sufficient evidence.

EKC studies in which ${{\text{CH}}}_{4}$ emissions are determined as pollutants are relatively few Skaza and Blais (2013), Madaleno and Moutinho (2021), Chaudhry et al. (2022). Cho et al. (2014) discussed the relationship between different GHG emissions components representing economic growth, energy use, electricity production and environmental degradation between 2008 and 2018 for 27 EU countries within the scope of EKC. The findings show that the EKC hypothesis is valid only for ${{\text{CH}}}_{4}$ emissions and ${{\text{NO}}}_{2}$ emissions in EU 12 group countries, and there is not enough evidence for GHG emissions and ${{\text{NO}}}_{2}$. Adeel-Farooq et al. (2020) examines the impact of economic growth, trade openness, energy consumption and financial development on ${{\text{CH}}}_{4}$ emissions in six ASEAN countries from 1985 to 2012. The findings show that trade openness does not have a significant effect on ${{\text{CH}}}_{4}$ emissions while confirm the inverted U-shaped relationship. Hassan and Nosheen (2019) examine the relationship between the three main pollution emissions (${{\text{CO}}}_{2}$ emissions, ${{\text{NO}}}_{2}$ emissions and ${{\text{CH}}}_{4}$ emissions) per capita income and rail transport in the context of the EKC hypothesis for 37 high-income countries between 1990 and 2017. The findings show that there is a U-shaped relationship for ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions, and an inverted U-shaped relationship for nitrogen emissions. Uddin (2020) examines the relationship between pollutants such as ${{\text{CO}}}_{2}$ emissions, pm25 and ${{\text{CH}}}_{4}$ emissions for 115 countries, agriculture and manufacturing GDP, urbanization, energy consumption, trade openness, transportation within the scope of EKC. The findings show that there is an inverted U-shaped relationship between ${{\text{CH}}}_{4}$ emissions and income for low, lower-middle and high-income countries, while trade openness has a reducing effect on ${{\text{CH}}}_{4}$ emissions in these country groups. Empirical findings from the above-mentioned studies indicate that there is no consensus on the validity of the EKC hypothesis.

In addition, many studies in the literature examine the relationship between human capital level and pollutant emissions. Halliru et al. (2020) discusses the relationship between human capital, financial development, trade openness, biocapacity and carbon emissions for 6 West African (ECOWAS) countries in the framework of EKC during the period 1970–2017. The results obtained here show that there is not enough evidence for the validity of the EKC, as well as that human capital has increased ${{\text{CO}}}_{2}$ emissions contrary to expectations. Tenaw and Beyene (2021) used different pollution indicators for the years 1990–2015 in 20 Sub-Saharan African countries and obtained sufficient evidence about the validity of the human capital-modified EKC in the long run. Hao et al. (2021), Rahman et al. (2021), Shafiullah et al. (2021) and Sheraz et al. (2021) found that human capital reduces carbon emissions. In addition, Hanif et al. (2020) investigated the relationship between renewable and non-renewable energy types and human capital under the EKC hypothesis for 30 developing countries in 1990–2017. Findings from this study, supporting the modified EKC, show that the development in human capital in the beginning increases non-renewable energy consumption, but when the development in human capital exceeds a certain turnin point, countries reduce their non-renewable energy consumption.

In the literature, there are many studies that examine environmental pollution by including the effect of financial development. For example Tamazian et al. (2009) examined the effects of economic growth, energy consumption, R&D expenditures and financial development on ${{\text{CO}}}_{2}$ emissions in BRICS countries in 1992–2004 in their study, which deals with many factors. They found that both of economic and financial development are determinants of the environmental quality in BRIC countries. In addition, Al-Mulali et al. (2015) investigated the effects of financial development on ${{\text{CO}}}_{2}$ emissions, along with variables such as urbanization, economic growth, trade openness, petroleum consumption, in his study of 129 countries for the period 1980–2011. The findings show that there is a cointegration relationship between the variables, and according to the dols and causality results, financial development will improve the environmental quality in the short and long term. Finally, Ganda (2019) examines the environmental effects of financial development in OECD countries through carbon emissions and ghg emissions in his study, which uses various financial indicators between 2001 and 2012. While the empirical findings show that domestic credit to the private sector are negatively related to environmental pollution, and foreign direct investment is positively related, the turning points obtained confirm the validity of the Kuznets curve hypothesis. In addition, Omri et al. (2015), Khaskheli et al. (2021), Dogan and Seker (2016), Park et al. (2018), Maneejuk et al. (2020) studies also address the relationship between financial development and environmental pollution from various aspects and present findings from the country groups discussed.

Since technological innovation will provide great gains in reducing ${{\text{CO}}}_{2}$ emissions, it can be considered as an effective factor for determining the relationship between economic growth and environmental pollution (Fethi & Rahuma, 2020; Nikzad & Sedigh, 2017, Shahzad et al. 2020). Based on empirical evidence supporting that countries technological improvement is a measure of their innovation capability,as a result, innovation and economic development are inseparably linked (Ding et al., 2021; Liu et al., 2017; Wang et al., 2021). For instance, Ahmed et al. (2016) analyzes the impact of technological innovation on environmental pollution employing 1980–2010 data for 24 European countries under the EKC hypothesis. The main results indicate that innovation has an effect on reducing environmental pollution. The results obtained from the study of Allard et al. (2018), in which they discuss the effect of innovation on environmental pollution with 1994–2012 data for 74 countries, show that innovation increases ${{\text{CO}}}_{2}$ emissions contrary to expectations. They think that this is due to the inclusion of only a variable that represents all of the technological developments in the model, rather than environmentally focused technological developments. And also the environmental-oriented technological developments factor is recently included in the EKC framework through the studies of Zhao et al. (2022), Wei and Lihua (2022) and Ahmad et al. (2021), Zhao et al. (2022) examines the relationship between carbon emissions, solar energy production and GDP in the G7 countries for the years 1995–2018 within the framework of the EKC hypothesis. The results show that solar energy and eco-innovation have a statistically significant and reducing effect on carbon dioxide emissions in the long term and the EKC is valid. On the other hand Khattak et al. (2020), investegated the relationship between innovation, renewable energy consumption, income and ${{\text{CO}}}_{2}$ emission within the framework of EKC hypothesis for BRICS economies in 1980–2016. The long-run coefficients obtained show that economic growth initially degrades the environment but later improves it in line with EKC’s hypothesis, however, innovation do not reduce the ${{\text{CO}}}_{2}$ emissions. For selected ASEAN countries, Wei and Lihua (2022) discuss the relationship between eco-innovation, economic growth, tourism and environmental pollution between 1995 and 2018 within the scope of the EKC hypothesis. The findings confirm the inverted U-shaped relationship between environmental pollution and economic growth and reveal that eco-innovation has a positive effect on environmental quality.

As seen in the literature, studies with different techniques used have dealt with EKC depending on income level. Moreover, although there are studies investigating the impact of human capital on polluting emissions, there are few studies addressing the moderator role in EKC. However, when the literature is examined, there are not enough studies on the role of human capital in the EKC hypothesis and the relationship between factors such as technological innovation and ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions. In this sense, this study aims to fill the gap in the literature.

3 Methodology, data and econometric model

3.1 Methodology

In this study, fixed effects and random effects panel data model are used for the empirical analysis. Equation 1 shows the general panel data regression model in which the independent variables are expressed as $x$, and the coefficients as $\alpha$ and $\beta$, with $i$ being the unit $t$ time dimension.

$$y_{it} = \alpha + \sum_{k = 1}^{K} \beta_{k} x_{itk} + \varepsilon_{it} ,\,\, i \in \left\{ {1,2, \ldots ,N} \right\},\,\,t \in \left\{ {1,2, \ldots ,T} \right\}$$

(1)

The assumptions of the error term in the panel data model, where ${\varepsilon }_{it}$ is the error term, present the fixed effects and random effects estimation procedure. These models are expressed as follows, respectively.

$$y_{it} = \alpha + \sum_{k = 1}^{K} \beta_{k} x_{itk} + \mu_{i} + \varepsilon_{it} ,\,\,i \in \left\{ {1,2, \ldots ,N} \right\},\,\,t \in \left\{ {1,2, \ldots ,T} \right\}$$

(2)

$$y_{{it}} = \alpha + \sum\limits_{{k = 1}}^{K} {\beta _{k} } x_{{itk}} + + v_{{it}} ,\;i \in \left\{ {1,2, \ldots ,N} \right\},t \in \left\{ {1,2, \ldots ,T} \right\}$$

(3)

$v_{it} = \varepsilon_{it} + \mu_{i}$

The Hausman specification test is adopted to choose the appropriate model between random effects or fixed effects for panel data analysis (Greene, 2003, Wooldridge, 2002). The specification test proposed in Hausman (1978) are routinely used as pretests in applied work.

$$H = \left( {\hat{\beta }_{{{\text{FE}}}} - \hat{\beta }_{{{\text{FE}}}} } \right)^{T} \left( {{\text{var}} \left( {\hat{\beta }_{{{\text{FE}}}} } \right) - {\text{var}} \left( {\hat{\beta }_{{{\text{RE}}}} } \right)} \right)^{ - 1} \left( {\hat{\beta }_{{{\text{FE}}}} - \hat{\beta }_{{{\text{RE}}}} } \right)$$

(4)

Here $\hat{\beta }_{{{\text{FE}}}}$ and $\hat{\beta }_{{{\text{RE}}}}$ denote kx1 vectors of estimated coefficients of random effect and fixed effect model. The null hypothesis underlying the Hausman Test’s is that the unobserved factors is not correlated with the regressors.

Additionally we use different estimation method for analyzing validity of EKC hyphothesis; such as panel-corrected standard errors (PCSE) by Beck and Katz (1995) and also been applied (D–K) by Driscoll–Kraay (1998) standard error method aim of confirm the estimation result. The D–K estimator, applies the Newey-West correction to the series of cross-sectional means to obtain robust standard errors from the pooled OLS model. Unlike standard techniques, the D–K (1998) method takes into account cross-sectional or spatial dependence, (Baloch et al., 2019, Sarkodie & Strezov, 2019), thus the dependency issue is eliminating by providing consistent and robust estimates of the standard errors (Driscoll–Kraay, 1998; Hoechle, 2007). In addition, PCSE estimation method is useful for obtaining robust long-term forecasts by taking into account cross-sectional dependence, heteroskedasticity, and serial correlation (Hoechle, 2007). On the otherhand Beck and Katz (1995) as an alternative to Parks (1967) and Kmenta (1986), propose replacing the panel corrected standard errors (PCSE) with OLS standard errors, thus suggesting that the PCSE estimator is robust according to Monte Carlo simulations.

3.2 Data and econometric model

In this paper we collect panel data over the 1993–2018 period for 39 countries and we investigated the relationship between GHG emissions such as ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions and human development. As seen in the literature summary discussed in the second chapter, the effect of income on environmental degradation has been discussed together with various variables in different studies. Previous studies in the literature as, Jahanger (2022), Zhang et al. (2021), Zia et al. (2021) provide limited empirical evidence for an EKC hypothesis focused on human development. Most notably, the sustainability perspective in the context of human development has not been discussed within the framework of the EKC hypothesis with different environmental pollution indicators. The major issue is to decide the relationship between the level of human development required for the reduction of environmental pollution. For the purpose of testing to applicability of the revisiting EKC hypothesis, we have modified the traditional EKC model, proxy as HDI insead of GDP, following Hussain and Dey (2021) and Tenaw and Beyene (2021). Thus, the form of the model for the determinants of ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions is defined as follows:

$${\text{EP}} = f\left( {{\text{HDI}},{\text{HDI}}^{2} ,{\text{EC}},{\text{POP}},{\text{FD}},{\text{PAT}}} \right)$$

(5)

where EP stands for environmental pollution and consists of two emission classes (${{\text{CO}}}_{2}$ emissions and ${{\text{CH}}}_{4}$ emissions), ${\text{HDI}}$ denotes Human Development Index, $EC$ denotes energy consumption, ${\text{POP}}$ is total population, ${\text{FD}}$ Financial Development and ${\text{PAT}}$ Environmental related patents. ${{\text{CO}}}_{2}$ emissions are measured in metric tons per capita, ${{\text{CH}}}_{4}$ emissions in kilotons of ${{\text{CO}}}_{2}$ equivalent, ${\text{EC}}$ in kg of oil equivalent per capita, ${\text{POP}}$ is measured as population total and ${\text{FD}}$ is measured as broad-based index of financial depth and ${\text{PAT}}$ measured as total patents of environmental-related technologies.

$${\text{CO}}_{2it} = \beta_{0} + \beta_{1} {\text{HDI}}_{it} + \beta_{2} {\text{HDI}}_{it}^{2} + \beta_{3} {\text{EC}}_{it} + \beta_{4} {\text{POP}}_{it} + \beta_{5} {\text{FD}}_{it} + \beta_{6} {\text{PAT}}_{it} + \varepsilon_{it}$$

(6)

$${\text{CH}}_{4it} = \beta_{0} + \beta_{1} {\text{HDI}}_{it} + \beta_{2} {\text{HDI}}_{it}^{2} + \beta_{3} {\text{EC}}_{it} + \beta_{4} {\text{POP}}_{it} + \beta_{5} {\text{FD}}_{it} + \beta_{6} {\text{PAT}}_{it} + \varepsilon_{it}$$

(7)

To represent the relationship, an alternative model is employed with the ${\text{HDI}}$ perspective, and two different pollution indicators are used to model environmental degradation. The Eqs. 6 and 7 provide the model specifications for ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions, respectively. As Begum et al. (2015) and Tenaw and Beyene (2021), we conduct an additionally test for joint significance of ${\text{HDI}}$ and ${{\text{HDI}}}^{2}$ and utilize the U-test approach by Lind and Mehlum (2010), to check an inverted u shape relationship of EKC. However, if the true relationship is monotonic, an U test can be applied to confirm the presence of an extremum by evaluating the slopes at the maximum and minimum values of the independent variable (Kacprzyk & Kuchta, 2020). Lind and Mehlum (2010) argued the inadequacy of standard econometric methods used in detecting the inverted U-shaped relationship and proposed the U-test, which provides a confidence interval for the turning point. Thus, we benefit from the U-test for further robustness in our findings from PCSE and D–K estimator. In the selection of control variables, we proceed in line with the theoretical frameworks highlighted in the relevant literature and focus on the determinants of emissions. Following Usman and Hammar (2021), Awan and Azam (2021) and Khan et al. (2023), we include control variables namely of financial development and environmental-related patents. Moreover, many studies indicate that energy consumption (Raza and Shah (2018), Amri (2018), Javid and Sharif (2016), Özokcu and Özdemir (2017) and population (Wang et al. (2015), Saqib and Benhmad (2021), Lin et al. (2016) are also major determinants of emissions. Therefore, energy consumption and population have been included to the regression model as control variables. The detailed description, abbreviation, and sources of the data are mentioned in Table 1, and also the detailed list of selected countries for the investigation are depicted in Appendix Table 9. Additionally, all variables taken in a natural logarithm in the models for control the country groups heterogeneity, and also sample period and countries were selected on the basis of data availability.

Table 1 Description of variable

In adition Table 2 presents a summary of statistics. We focus on the dependent variables and the main independent variables that are thought to have an effect on the dependent variable. According to the descriptive statistics in Table 1, while the average of ${{\text{CO}}}_{2}$ for the whole panel was approximately 7.8 million tons per capita, it took the values of minimum 0.70 and maximum 30 million tons per capita between 1993 and 2018 in these 39 countries. It has also been, the HDI ranges widely, with the minimum value of 1.55 and the maximum value of 4.15, indicating a rather heterogeneous dataset. This is not surprising as the countries in the sample include a large panel and offer a broad spectrum for the HDI around the world. Table 2 also shows that the average of per capita ${{\text{CO}}}_{2}$ and total ${{\text{CH}}}_{4}$ emissions, which are determined as different dependent variables, are 7.8224 and 229,428.2 respectively. In addition, the financial development average and Std. Dev. values are 0.5812 and 0.1982, respectively, showing that the fluctuation change is insignificant.

Table 2 Summary statistics

4 Empirical results and discussion

First, the presence of individual effects that could not be observed in the EKC models for ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions is determining by the F-test; this results are reported in the first row of Table 3. The null hypothesis that there is no individual effect was rejected which in the model determined for both pollution indicators. In the next step, the Hausman test is applied to determine whether the individual effects are related to the independent variables. The findings of the Hausman test indicate that the individual effect is correlated with the independent variables in the ${{\text{CO}}}_{2}$ emissions and ${{\text{CH}}}_{4}$ emissions models.

Table 3 Hausman test

The tests performed to test the existence of heteroscedasticity and autocorrelation in the models are shown in Table 4. Concerning the heteroscedasticity and autocorrelation, the results of Wald test shows presence of heteroscedasticity, and BW test result indicate that there is autocorrelation in both of the models.

Table 4 Heteroscedasticity, and autocorrelation tests

To detect the presence of cross section dependence, the Peserans’s (2004) cross-section dependence (CD) was applied for all variables. As shown in Table 5 The null hypothesis of cross-sectional independence was strongly rejected at a 1% level for all the variables.

Table 5 Tests for cross-section dependence

Then, moving on to inspect the stationary properties of the series. Harris and Tzavalis (1999) and Breitung (2001) unit root tests exhibit mixed result at levels, but all variables are found to be stationary at first difference in Table 10 (in Appendix). We also interest to expose the causal relationship between variables for the sample countries. To that end, we apply the bootstrap panel Granger causality test developed by Emirmahmutoglu and Kose (2011) instead of traditional causality methods because this approach considering the mixed order of integration the variables and accounts for both issues of heterogeneity and cross-sectional dependence. Results from the bootstrap causality test are summarized in Table 6.

Table 6 Causality test results

The findings show that there is a two-way causal relationship between ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions with human development in the panel of sample countries, respectively. There is also one-way causality running from financial development to energy consumption. We observe a two-way causal relationship between environmental-related technologies, population and energy consumption with both of ${{\text{CO}}}_{2}$ and ${{\text{CH}}}_{4}$ emissions, respectively. Furthermore, the results show that there is a two-way causal relationship between financial development and CH₄ emissions, but there is no causality between CO₂ emissions and financial development. We also observe two-way causal relationship between environmental-related technologies and human development, and between energy consumption and human development.

Afterward, we proceed to examine the relationships between human capital, energy consumption, population, financial development and environmental-related technologies with CO₂ and CH₄ emissons. Following the Hausman test results in Table 3, fixed effects estimator is adopted to examine the relationship between in Eqs. 6 and 7 (the results presented in Appendix Table 11). Besides, we employ the alternative estimation methods to the fixed-effects model such as the D–K and PCSE estimations methods to fight against heteroscedasticity, cross-sectional dependence and autocorrelation problem. The empirical findings of the two different environmental pollution indicators with CO₂ and CH₄ emissions, obtained the D–K and the PCSE estimators are summarized in Tables 7 and 8. Moreover, since traditional EKC analysis points to a long-run phenomenon for different periods of a growing economy over time, the focus of the article is to determine the role of HDI in the analysis of the modified EKC relationship. The model results constructed by different estimators in Table 7 shows that CO₂ emissions initially increase up to the turning point of HDI, but then when a certain turning point is reached, CO₂ emissions begin to decrease, revealing an inverted U-shaped relationship. In addition the coefficient estimates obtained for ${{\text{CH}}}_{4}$ emissions also point to similar findings and confirm the inverted U-shaped relationship between HDI and ${{\text{CH}}}_{4}$ emissions. Our parameter estimation results for ${\text{HDI}}$ and ${{\text{HDI}}}^{2}$ in Table 8, similar with Payab et al. (2023). However, most of the studies examining the EKC hypothesis consider sufficient conditions for the coefficients of the parabolic model to make statistical inferences about the existence of an inverted U-shaped (Sarkodie, 2018; Tenaw & Beyene, 2021). We report the test results for the appropriate test of inverted U-shaped relationship proposed by Lind and Mehlum (2010) in the last part of the Tables 7 and 8. Similar to the traditional EKC hypothesis presented by economic growth, the results of the modified EKC hypothesis substituted by HDI; HDI increases CO₂ and CH₄ emissions initially but reduces emissions after the turning point. The D–K and PCSE estimation results in Table 7 indicate that at the beginning CO₂ emission increased up to the HDI turning point of 0.8127 and 1.008 respectively, and then begins to reduce after crossing the turning point, supporting an inverted U-shaped relationship. In addition Fig. 1, provides additional line of visual evidence the modified EKC relationship between CO₂ emissions, CH₄ emissions and HDI based on turning points. In addition, the results of the U test developed by Lind and Mehlum (2010) demonstrates that rejects the monotonous or U-shaped composite null hypothesis in favor of an inverted U-shaped relationship. Contrary to the PCSE estimator in Table 8, the results for the DK estimator indicate that the necessary and sufficient conditions of the U test of Lind and Mehlum (2010) not verify the existence of an inverted U-shape relationship for the ${{\text{CH}}}_{4}$ emission. (see Fig. 1). Evidence from the results shows that the relationship between ${{\text{CH}}}_{4}$ emission and HDI in this panel country exhibits a monotonic trend with a slope coefficent bounds of 1.1876 and − 0.1058; but the turning point occurs at a 1.3438 and statistically insignificant. These findings shows between ${{\text{CH}}}_{4}$ emission and a HDI the effect of scale Stern (1998) that causes a monotonic increase. Consequently, the turning points obtained for human development, differ for difference classes of GHG emissions such as CO₂ and CH₄. In general, we observe the lower (0.8127, 1.0080) turning points for CO₂ emissions, while CH₄ produces the higher (1.3438, 1.1017) turning points.

Table 7 Estimation result of CO₂

Table 8 Estimation result of CH₄

Fig. 1

Illustration of U-shape relationship

Given the above findings effect of the PAT on CO₂ and CH₄ emissions is generally negative and statistically significant in Tables 7 and 8, hence growth in PAT will improve environmental quality. D–K estimations obtained from Eq. 6 indicate that a 1% increase in PAT decreases CO₂ emission by 0.025%. And also all estimation from Eq. 7 with CH₄ emission model support that this results. In addition, while the effect of PAT on CH₄ emissions is statistically significant and negative, and also we can see that with all estimators 1% increase in PAT decreases CH₄ emission by 0.06% and 0.047%, respectively. The encouraging role of PAT in these panel data countries, as similar to Zhang et al. (2017), Allard et al. (2018), Tao et al. (2021) indicates that environmental innovations, effectively tackling environmental problems and promoting suistainnable development. Besides, considered that environmental pressure resulting from, population growth and increased energy consumpiton would further encourage these countries adopt environmental benefit and pursue alternative energy sources by improving PAT investments.

On the other hand, positive and statistically significant coefficient estimates are found for FD (Eq. 6), POP and EC (Eqs. 6 and 7). The results for POP shows generaly a significant positive impact on the CO₂ and CH₄ emissions because these pollutuants in the atmosphere are the result of different kind of human activities (Sohag et al., 2017; Anser, 2019). These findigs indicate that POP increas the environmental pollution, provoking the CO₂ and CH₄ emissions for the panel countries. According to the findings of the Table 7, a 1% increase in POP increases CO₂ emissions by 0.034%. According to the DK estimates from Eq. 7 presented in Table 8, a 1% increase in POP reduces CH₄ emissions by 0.211, besides that according to PCSE estimations 1% increase in POP increases CH₄ emissions by 1.026%. Anser et al. (2020), Zoundi(2017) Acheampong et al. (2019) confirmed similar evidence for the significant influence of POP in dealing with environmental pollution. Moreover same is the case with coefficients estimates of energy consumption. For estimations by D–K and PCSE estimator, every 1% increase in EC increases CO₂ emissions by 1.067 and 0.799 respectively. Similarly, 1% increase in EC increases CH₄ emissions by 0.663, and 0.196 respectively (Tables 7, 8). Note that the coefficient estimates for POP and EC are expected positive and are coherent with findings in the empirical literature Dong et al. (2019), Wang and Dong (2019), Behera and Dash (2017). Consequently, these countries require more energy for economic growth but the lack of focus on renewable energy sources should not be an option to spur sustainable growth trajectories.

Interestingly, empirical results from the DK estimator present the asymmetric nature of financial development on kind of emissions. The result from financial development positively affects CO₂ emissions. In other words, shows that from the DK estimator 1% increases in FD causes 0.063% growth in CO₂ emissions (Table 7). The findings are consistent with Le and Ozturk (2020) Tamazian and Rao (2010), Saidi and Hammami (2015) and Dar and Asif (2018). And also result of the different estimation methods in Table 8 shows that a 1% increase in FD decreases CH₄ emissions by 0.332% and 0.158% respectively. These results for the CH₄ emissions model are in line with the recent studies of Al-mulali et al. (2015), Khan et al. (2019), Usman et al. (2022), conducted in the finance-emissions context.

5 Conclusion and discussion

This study inspects the environment-development link within a sustainability-focused EKC framework in a panel of 39 selected countries using data from 1993–2018. To do so, we modified the traditional EKC hypothesis with human development and estimate turning points for a pollutant emissions (i.e., CO₂; CH₄). The study differs in that it employs a U-test of Lind and Mehlum (2010) for an inverted U-shape which is usually neglected in previous studies, and also focuses on human development on the modified Kuznets curve hypothesis as an indicator of the economic development. It is important to emphasize that the findings are limited to the panel data countries included in the analysis due to data availability. At this point, next researches should include different economies by changing the type of data period or explanatory variables. The findings of this study show an inverted U-shaped relationship between human development and CO₂ emissions, confirming the validity of the Environmental Kuznets Curve (EKC) hypothesis. In addition, the findings in this paper provide empirical evidence of the negative effects of increased energy consumption on CO₂ and CH₄ emissions, while demonstrating that human capital reduces emission levels. In contrast, there is a monotonically increasing relationship for CH₄ emissions in panel countries for the D–K estimators unlike the PCSE estimations, invalidating the EKC hypothesis. Obviously, the incremental human development, financial development, or energy consumpiton enhancement can encourage the economic activities and, hereby extend the scale effects of economic activities on CH₄ emissions.

From the empirical findings of this study remarkable suggestions can be made for policy makers to take into account improvements with the focus on sustainable energy consumption, environmental innovation, financial development, human development, and environmental quality. Thus, we emphasize the several policy implications made on the bases of the empirical results of the study. Primarily, as a reflection of the validity of a modified EKC hypothesis, economic development can be viewed as a key of the environmental issue. In other words, It will require a lot of effort to achieve environmental sustainability in the early stages of economic development, nevertheless environmental progress may be possible with the increase in the human development levels of the countries. For this reason, countries should include practices that encourage the protection or increase of the level of human development in planning in order to provide a sustainable environmental policy. This reflects that countries do not need to compromise on growth to control environmental degradation. Accelerating the development of human capital enables individuals to be aware of the significance of environmental quality; moreover, it results in an environmentally friendly, educated, healthy, and productive society. Additionally, encouraging quality education opportunities that will boost research and development activities for clean technologies and sustainable environmental quality is crucial Thus, governments should invest intensively in human development for the future of environmental quality. The improvement of environmental innovation in the countries interacts with CO₂ emissions and CH₄ emission. These findings mean that it is possible and doable to impact positively the overall environmental quality by encouraging applications for environmentally related patents. Additionally, achieving the reduction of greenhouse gas emissions, widely recognized as an indicator of sustainable development, can be facilitated through subsidies, investments, and tax incentives specifically targeted for the area of technological innovation. In this context, we recommend that governments support markets through regulations that encourage environmentally focused technological innovations. It is important to adopt that improving environmental sustainability, such as new environmentally-oriented technologies and other environmentally friendly growth options in this perspective. Moreover, the results show that consistence with recent studies in literature on the relationships between innovation, financial development and environmental pollution and provide policymakers with remarkable evidence of the precise effects of innovation and financial development on environmental pollution. On the other hand, concerning to energy consumption, the findings show that policy makers should have various environmental policies such as, increasing the efficiency of old-fashioned technologies, promoting effective ways of using renewable energy, developing environmentally oriented technological improvements for sustainable development. Policymakers could also encourage research and development programs in the area of clean and sustainable energies.

Appendices

Appendix 1

See Figs.

Fig. 2

${{\text{CH}}}_{4}$ emission by country for 2018

2 and

Fig. 3

${{\text{CO}}}_{2}$ emission by country for 2018

3.

Appendix 2

See Tables 

Table 9 List of countries included in the dataset

9,

Table 10 Panel unit root tests result

10 and

Table 11 FE model estimation results for CO₂ and CH₄ emissions

11.