The Effect of Renewable Energy Use and ICT Development on CO₂ Emissions in EU Transition Economies: Evidence from Causality and Cointegration Analyses Under the Presence of Cross-Sectional Dependence and Heterogeneity

Abstract

CO₂ emissions are amongst the most significant contributors to global warming and climate change and continue to increase throughout the world. In this regard, this study investigates the interplay amongst renewable energy use, ICTs, economic development, and CO₂ emissions in EU transition economies during the years of 2000–2021 through Emirmahmutoglu and Kose’s causality approach and LM bootstrap cointegration test. Panel-level causality analysis indicates a feedback interaction amongst renewable energy use, economic development, and CO₂ emissions, but a one-way causal effect of CO₂ emissions on ICT development. However, country-level causality analysis shows that the causal relationships amongst renewable energy use, ICTs, economic development, and CO₂ emissions change among EU transition economies. The estimated cointegration coefficients reveal that renewable energy use has a negative impact on CO₂ emissions in all countries, while the effects of ICTs and economic development on CO₂ emissions differ amongst the countries. The findings of this study emphasize the significant roles of renewable energy use and ICTs to reduce CO₂ emissions.

1. Introduction and Background

World-wide climate change and global warming are major challenges, and CO₂ emissions are a key factor behind global warming and climate change. However, world-wide CO₂ emissions have considerably risen, mainly from fossil fuel use, industrial emissions, deforestation, and agriculture [1]. In this context, China, the United States, India, the EU27, Russia, and Brazil were the largest CO₂ emitters in 2023 [2]. Therefore, international organizations such as the United Nations Environment Programme, the World Nature Organization, and countries, especially developed countries and the EU (European Union) have been in search of measures to decrease CO₂ emissions.

Furthermore, one of the main goals of the Millennium Development Goals and Sustainable Development Goals (SDGs), which were set by the UN (United Nations) members in 2000 and 2015, respectively, is to achieve environmental sustainability [3,4]. In this regard, CO₂ emissions, accounting for over 70% of world-wide greenhouse gas emissions, should be decreased by 45% by 2030 compared to 2010 levels, and countries must attain net-zero emissions by 2050 to make progress in environmental sustainability and limit global warming in the context of the SDGs and the Paris Agreement [5].

In the associated literature, extensive numbers of demographical, institutional, political, socio-economic, technological, and energy-related factors have been suggested as drivers of rising global CO₂ emissions [6,7,8,9]. This research concentrates on the interaction between renewable energy (RNW) use, ICTs (information and communication technologies), economic development, and CO₂ emissions in EU transition economies, because these countries have come a long way in ICTs, RNW use, and economic development since the 2000s. Furthermore, the EU has followed a stringent environmental policy and energy policy to enhance the share of RNW use.

Renewable sources such as wind, sunlight, and biomass are unlimited and naturally renew themselves, and RNW use emits much lower levels of harmful greenhouse gases when compared with fossil fuels [10]. Therefore, adopting clean energy production such as RNW is required to reach net-zero emissions and make progress in environmental sustainability. However, the production of renewable energy necessitates a significant amount of water and land, and in turn, these renewable energy sources can negatively impact environmental quality through an increasing ecological footprint [11]. In conclusion, the nexus between RNW use and CO₂ emissions can differ based on these views.

In the literature, four theories including (1) first-, second-, and third-order effects, (2) ecological world systems theory (EWST), theory of technological determinism, and (3) socio-technical systems theory (STST) have been developed to explain the nexus between ICTs and the environment [12]. In this regard, the first-order effects occur, because increasing energy consumption during the lifespan of ICT products will foster environmental harm [13]. On the other hand, the second-order effects refer to the negative relationship between ICTs and environmental harm due to the ICT-sourced improvements in energy efficiency, resource management, and digitalization [14,15,16]. The third-order effects (also known as rebound effects or Jevons paradox) occur when the negative environmental effects of ICTs outweigh the environmental benefits of ICTs due to the increases in ICT products’ demand [17].

Furthermore, the EWST of Hornborg [18] suggests that improvements in one component of the global system are offset by breakups in another component. In this case, the periphery is exposed to environmental and social problems when the core employs natural resources to increase welfare through technological progress [19]. The theory of technological determinism states that ICTs foster environmental quality through improvements in efficiency and real-time monitoring [20], and the STST asserts that ICTs also impact environment through the interaction between technological progress and social structures [21]. ICTs are also a significant driver of economic growth [22,23], and in turn, the growth influence of ICTs on CO₂ emissions can be changed within the scope of the EKC hypothesis. Lastly, on the one hand, ICTs can contribute to environmental protection through improving the optimization of RNW sources [24]. On the other hand, increases in RNW use can decrease the negative environmental effects resulting from ICTs’ increasing energy use. Consequently, the nexus between ICTs and CO₂ emissions can be varied depending on the above-mentioned theoretical views.

Economic development is generally substituted by per capita GDP in the literature, but this study utilizes the concept of human development represented by the human development index (HDI), a composite indicator of education (mean and expected schooling years), health (life expectancy at birth), and living standards (per capita gross national income) [25]. The connection between economic development and CO₂ emissions can be explained by the EKC hypothesis, which proposes an inverted U-shaped association amidst economic development and environmental harm. In this sense, environmental harm occurs at the initial stages of development, but environmental harm begins to decrease after a threshold development level depending on countries’ own socio-economic characteristics [26]. On the one hand, economic development can impact environmental harm through economic growth and development due to increases in energy use, transportation, production, and consumption. Furthermore, improvements in economic development cause individuals to better understand the implications of environmental harm and how to combat it [27]. We expect a significant causal interaction between CO₂ emissions and economic development based on these considerations, but the direction of the causality may be different based on the socio-economic structure of the countries.

This study explores the effect of RNW use, ICTs, and human development on CO₂ emissions in the EU transition countries. These countries have increased the share of RNW use in total energy use and achieved a remarkable improvement in human development [28,29,30]. The European Green Deal of 2019 initiated by the EU Commission to prioritize environmental concerns has also played a significant role in the growing RNW use [31,32]. Therefore, this study is significant in its determination of the effect of the EU’s environmental and energy policies on CO₂ emissions. Furthermore, the results of the study would be useful to design EU policies related to ICTs and human development by uncovering the environmental effects of remarkable improvement in ICTs and human development.

In the associated empirical literature, academics have usually focused on the influence of CO₂ emissions on human development, but the effect of human development on CO₂ emissions has been analyzed only by Akbar et al. [33] and Alkan et al. [34]. Therefore, this study would also be one of the first empirical studies examining the effect of human development on CO₂ emissions. Furthermore, the related theoretical and empirical literature on the nexus between ICTs and the environment remains inconclusive, and further empirical studies like our study would be useful to understand the multifaceted interaction between ICTs and the environment. Last, the study shows the environmental effects of the RNW energy transition by the EU transition economies. The next section of the article introduces a review of the literature on human development, RNW use, ICT development, and CO₂ emissions, and Section 3 explains the data and methodological approach of this research. Section 4 describes the econometric tests and discusses their outcomes, and the conclusions are presented in Section 5.

2. Literature Overview

The growing CO₂ emissions and the negative environmental and health effects resulting from these increases have caused academics to research the driving factors of global CO₂ emissions. In this connection, this study investigates the interaction nexus between RNW use, ICTs, economic development, and CO₂ emissions.

The negative environmental and health implications of fossil fuels, volatility in fossil fuels’ prices, disruptions in fossil fuels’ supply, and the growing concerns related to energy security have encouraged countries to transition towards RNW production. In this connection, the environmental effects of RNW use have been explored by a vast number of researchers, and almost all of them have uncovered a negative influence of RNW use on CO₂ as introduced in

Table 1. Summary of the recent literature on the nexus between RNW and CO₂ emissions.

Studies	Countries; Analysis Duration	Method	Impact of RNW on CO2 Emissions; Causality
Otim et al. [9]	East African Community countries; 1996–2019	Cointegration and causality tests	Negative; unilateral causality from CO2 emissions to RNW energy use
Pal et al. [35]	Asian states; 1997–2023	Panel ARDL	Negative
Dogan and Seker [36]	EU-15; 1980–2012	Cointegration and causality	Negative; bidirectional causality
Inglesi-Lotz and Dogan [37]	Sub-Saharan African countries; 1980–2011	Causality test	Causality from RNW to CO2 emissions
Chen et al. [38]	China; 1980–2014	ARDL	Negative; bidirectional causality
Ben Jebli et al. [39]	120 countries; 1990–2015	Regression	Negative
Saidi and Omri [40]	15 major RNW users; 1990–2014	Cointegration and causality tests	Negative; bidirectional causality
Shahnazi and Dehghan Shabani [41]	EU members; 2000–2017	Regression	Negative
Szetela et al. [42]	Top natural resource-dependent economies; 2000–2015	Regression	Negative
Kuldasheva and Salahodjaev [43]	Rapidly urbanizing countries; 2000–2015	Regression	Negative
Mukhtarov et al. [44]	Azerbaijan; 1993–2019	Cointegration	Negative
Guo et al. [45]	42 countries; 2000–2014	Regression	Negative
Justice et al. [46]	Ghana; 1990–2020	Regression	Negative
Sezgin et al. [47]	BRICS states; 2000–2020	Causality and cointegration tests	Negative; bidirectional causality
Almulhim et al. [48]	BRICS states; 1996–2020	MMQR	Negative
Lorente-de-Las-Casas and Marrero [49]	OECD members; 1990–2019	Regression	Negative

[9,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52]. However, only Otim et al. [9], Dogan and Seker [36], Inglesi-Lotz and Dogan [37], Chen et al. [38], Saidi and Omri [40], and Sezgin et al. [47] analyzed the causal interplay between RNW use and CO₂ emissions. Dogan and Seker [36], Chen et al. [38], Saidi and Omri [40], Sezgin et al. [47] uncovered a bidirectional causality between two series, but Inglesi-Lotz and Dogan [37] unveiled a unilateral causality from RNW use to CO₂ emissions, while Otim et al. [9] discovered a causality from CO₂ emissions to RNW use.

The environmental effects of ICTs can change depending on the utilization of ICTs and the socio-economic development level of a country. Hence, the related literature summary introduced in

Table 2. The recent literature on the relation between ICT indicators and CO₂ emissions.

Studies	Countries; Analysis Duration	Method	Impact of Diverse ICT Indicators on CO2 Emissions; Causality
Nguyen et al. [50]	Selected G20 countries; 2000–2014	FMOLS and quantile regression	Negative
Wen et al. [51]	MINT countries; 1990–2018	Regression	Negative
Chang et al. [52]	10 developed countries; 1990–2019	ARDL	Negative
Islam et al. [53]	GCC countries; 1995–2019	Cointegration and causality tests	Negative; bidirectional causality
Linghu et al. [54]	30 Chinese provinces; 2008–2019	Regression	Negative
Park et al. [55]	Selected EU members; 2001–2014	Cointegration and causality tests	Positive; a unilateral causal relation from internet usage to CO2 emissions
Raheem et al. [56]	G7 countries; 1990–2014	Cointegration	Positive
Chatti [57]	43 countries; 2022–2014	Regression	Positive
Ebaidalla and Abusin [58]	GCC countries; 1995–2018	MG and AMG estimators	Positive
Kim [59]	OECD members; 1990–2018	PMG estimator	Positive
Al-Mulali et al. [60]	77 countries; 2000–2013	Regression	Negative (developed countries); insignificant (developing countries)
Appiah-Otoo et al. [61]	110 countries; 2000–2018	Regression	Negative (countries having high ICT quality); positive (countries having moderate and low ICT quality.)
Shabani and Shahnazi [62]	Iran; 2002–2013	Cointegration and causality tests	Positive (industrial sector); negative (service and transportation sectors); bidirectional causality (transportation and industrial sectors); unidirectional causality from ICT to CO2 emissions (service sector)
Altinoz et al. [63]	China, India, Mexico, Brazil, Turkey, Thailand, South Africa, Malaysia, Russia, and Indonesia; 1995–2014	VAR	Fixed broadband subscriptions and internet usage (+); mobile cellular subscriptions (−); bidirectional causality
Li et al. [15]	91 countries; 2007–2015		Positive (BRI countries) and inverted U-relation (non-BRI countries)
Khan et al. [64]	91 countries; 1990–2017	Regression	Inverted-U relationship; negative (developed countries) and positive (developing countries)
Amri et al. [65]	Tunisia; 1975–2014	ARDL	Insignificant

supports these theoretical views. On the one hand, Nguyen et al. [50], Wen et al. [51], Chang et al. [52], Islam et al. [53], and Linghu et al. [54] disclosed a negative influence of ICT indicators on CO₂ emissions. However, the other studies found a positive association between different ICT indicators and CO₂ emissions [55,56,57,58,59]. Accordingly, Al-Mulali et al. [60], Appiah-Otoo et al. [61], Shabani [62], and Altinoz et al. [63] showed mixed consequences of the relation between diverse ICT indicators and CO₂ emissions. Furthermore, Li et al. [15] and Khan et al. [64] discovered an inverted-U relation between diverse ICT indicators and CO₂ emissions, while Amri et al. [65] revealed an insignificant relation between the two indicators.

However, only Islam et al. [53], Park et al. [55], Shabani and Shahnazi [62] and Altinoz et al. [63] researched the causal relation between diverse ICT indicators and CO₂ emissions, and Islam et al. [53], Shabani and Shahnazi [62], and Altinoz et al. [63] discovered a bidirectional causal relation between the two series, while Park et al. [55] revealed a unilateral causal relation from internet usage to CO₂ emissions.

In the empirical literature, academics have usually concentrated on the influence of CO₂ emissions on human development. In this regard, only Akbar et al. [33] and Alkan et al. [34] analyzed the effect of human development on CO₂ emissions. Akbar et al. [33] investigated the relation between health expenditures, human development, and CO₂ emissions in the OECD members between 2006 and 2016 by way of the VAR method and discovered a negative influence of the human development index on CO₂ emissions. On the other hand, Alkan et al. [34] also investigated the relationship between human development and CO₂ emissions in selected country groups between 1994 and 2021 via the cointegration test and unveiled a positive influence of human development on CO₂ emissions.

Furthermore, Bedir and Yilmaz [66], Sezgin et al. [67], Alola et al. [68], and Minh and Ly [69] performed a two-way analysis on the nexus between CO₂ emissions and human development but uncovered different causalities between the two variables for different countries.

Bedir and Yilmaz [66] studied the causal relation between the HDI and CO₂ emissions in OECD members between 1992 and 2011 by way of the Kónya causality test; their outcomes showed unidirectional causality from CO₂ emissions to the human development index in the USA, Turkey, Spain, Polan, Korea, Japan, and Italy, one-way causality from HDI to CO₂ emissions in Mexico and France, and bidirectional causality in Switzerland and Norway.

Sezgin et al. [67] examined the interplay between environment policies, HDI, and CO₂ emissions between 1995 and 2015 in the BRICS and G7 countries by way of cointegration and causality methods, and their results demonstrated a bilateral causal relation between CO₂ emissions and HDI in the UK, the USA, Japan, and Germany and a unilateral causal nexus from CO₂ emissions to HDI in Brazil, Canada, China, and France. Furthermore, the cointegration coefficients unveiled a negative effect of HDI on CO₂ emissions.

Alola et al. [68] analyzed the interaction between human development and harmful emissions in the USA between 1990 and 2019 by way of a frequency domain Granger causality and disclosed a causality from carbon monoxide, sulfur dioxide, PM2.5, and PM10 to human development and a causality from human development to nitrogen oxides and sulfur dioxide. Lastly, Minh and Ly [69] researched the influence of human development on CO₂ emissions in Vietnam between 1990 and 2020 by way of a causality test and a VAR approach and unveiled insignificant causality between the two variables, but the results of the VAR demonstrated a positive influence of human development on CO₂ emissions.

On the other hand, the empirical studies exploring the effect of CO₂ emissions on human development have discovered different results. On the one hand, Asongu and Odhiambo [70] uncovered a negative effect of CO₂ emissions on human development. Nevertheless, Adekoya et al. [71], Sallam et al. [72], Ezako [73], and Ahmed and Alhassoon [74] found a positive effect of CO₂ emissions on human development. However, Fakhri et al. [75] uncovered that the relationship between human development and CO₂ emissions varied between countries. Furthermore, Akpolat and Bakırtaş [76] disclosed an inverted U-shaped relation between CO₂ emissions and human development. Last, Porcher et al. [77] also suggested that CO₂ emissions supported human development to a certain level and had no significant effects on human development after this threshold.

Asongu and Odhiambo [70] explored the influence of CO₂ emissions on human development in Sub-Saharan African states between 2000 and 2012 by way of regression and uncovered a negative influence of CO₂ emissions on human development. On the other hand, Adekoya et al. [71] studied the impact of CO₂ emissions on human development in 126 states between 2000 and 2014 by way of regression and uncovered a positive relationship between human development and CO₂ emissions.

In a similar vein, Sallam et al. [72] researched the influence of CO₂ emissions on the HDI in the MENA states between 1990 and 2018 employing regression and unveiled a positive influence of CO₂ emissions on the HDI. Ezako [73] also explored the relation between human development and CO₂ emissions in 56 developing economies between 2005 and 2019 by way of the ARDL and unveiled a positive influence of CO₂ emissions on human development. Ahmed and Alhassoon [74] studied the influence of CO₂ emissions on the HDI in Saudi Arabia between 1990 and 2021 via ARDL and discovered a positive influence of CO₂ on the HDI.

Fakhri et al. [75] explored the influence of CO₂ emissions on the HDI in the UK, Saudi Arabia, France, Morocco, Mexico, the USA, Sweden, and Norway between 1990 and 2021 through the ARDL approach and disclosed that CO₂ emissions negatively impacted human development in the UK and Sweden, positively impacted human development in Saudi Arabia and Norway, and had an insignificant effect on human development in the remaining states.

Akpolat and Bakırtaş [76] studied the effect of fossil energy, RNW, and CO₂ emissions on HDI in BRICS countries and Egypt, Iran, and Turkey via regression between 1990 and 2021 and revealed an inverted U-shaped impact on the HDI. Lastly, Porcher et al. [77] explored the influence of CO₂ emissions on human development in 119 states and uncovered that additional carbon consumption did not make a contribution to human development after a certain human development level.

3. Data and Methods

This research investigates the short and long-term interaction between RNW use, ICT development, economic development, and CO₂ emissions in the EU transition countries during the period of 2000–2021. In the applied part of the research, CO₂ emissions (COEM) are represented by CO₂ emissions per capita and are sourced from Climate Watch [78]. In addition, RNW use is substituted by renewable energy consumption in a country and sourced from the World Bank [28]. The ICT index of UNCTADSTAT [29] represents ICT development and is calculated by employing the usage of internet, fixed line, and mobile phones together with server security. Economic development (ECNDEV) is proxied by the HDI of the UNDP [25]. The variables utilized in the causality and cointegration analyses along with their sources are summarized in

Table 3. Dataset summary.

Variables	Explanation	Data Source
RNW	Renewable energy consumption (% of total final energy consumption)	World Bank [28]
ICT	ICT index	UNCTADSTAT [29]
ECNDEV	Human development index	UNDP [25]
COEM	CO2 emissions (tCO2e per capita)	Climate Watch [78]

.

The study’s dataset covers the years between 2000 and 2021. The beginning of the dataset is specified as 2000 due to the calculation of the ICT index as of 2000, and the end of the dataset is determined as 2021 because the data of RNW use ended in 2021. Stata 17.0 is utilized to perform CD, heterogeneity and unit root tests, and the AMG (augmented mean group) estimation of Eberhart and Bond [79]. Gauss 12.0 is used for the LM bootstrap cointegration test of Westerlund and Edgerton [80], and EViews 12.0 is employed to implement the E-K (Emirmahmutoglu and Kose) [81] causality test. The summary figures of COEM, RNW use, ICT, and ECNDEV are reported in

Table 4. Summary figures of COEM, RNW, ICT, and ECNDEV.

Variables	Mean Value	Standard Deviation	Minimum	Maximum
COEM	6.43	2.755	2.93	14.71
RNW	19.327	9.555	3.7	44
ICT	55.394	12.722	22.513	84.683
ECNDEV	0.843	0.041	0.73	0.924

. The average figures of COEM, RNW, ICT, and ECNDEV are, respectively, 6.43 metric tons, 0.843, 19.327%, and 55.394. HUMDEV shows a small variation among the EU transition countries, COEM and RNW use exhibit a moderate change among these countries, and ICT considerably varies between these countries.

The major goal of this research is to perform a short and long-term analysis of CO₂ emissions, human development, RNW use, and ICTs by way of the E-K causality test and the LM bootstrap cointegration test in comparison with the CD and heterogeneity characteristics of the study’s dataset.

The causal connections between two variables are analyzed by means of the Granger causality approach, which enables us to determine whether one variable is helpful in predicting another variable [82]. Furthermore, a causality test makes a two-way analysis between two variables possible, differently from the other econometric tools. On the other hand, cointegration method is used to identify whether two or more non-stationary series exhibit a co-movement over the long term [83].

The E-K causality test is derived from Toda–Yamamoto causality for panel datasets. This test has more information owing to the utilization of level values of the series and can be used for the series with I(0) or I(1) differently from the other panel causality tests [81]. The appropriate lag length (p) is firstly specified, and it is summed with the maximum integration level ($d_{max}$). Then, the panel VAR model introduced in Equations (1) and (2) is estimated by level values of the series for $p+d_{max}$ [81]:

$$x_{i,t}={\mu }_i^X\sum _{j=1}^{k_i+d_{{max}_i}}{A}_{11,i_j}{x}_{i,t - i}+\sum _{j=1}^{k_i+d_{{max}_i}}{A}_{12,i_j}{y}_{i,t - j}+u_{i,t}^x,$$

(1)

$$y_{i,t}={\mu }_i^y\sum _{j=1}^{k_i+d_{{max}_i}}{A}_{21,i_j}{x}_{i,t-i}+\sum _{j=1}^{k_i+d_{{max}_i}}{A}_{22,i_j}{y}_{i,t-j}+u_{i,t}^y,$$

(2)

where $y$ is COEM, and $x$ represents HUMDEV, RNW, and ICT.

The E-K causality test takes notice of both CD and heterogeneity among the cross sections.

Furthermore, the LM bootstrap cointegration test is utilized to analyze the long-term interaction between COEM, RNW use, ICTs, and ECNDEV, because the test takes account of CD and generates consistent results in small samples [80]. Finally, cointegration coefficients are predicted by means of AMG estimator, because this estimator results in robust findings in the presence of CD and heterogeneity [79].

4. Results

In this section, firstly, we present the results of the CD and heterogeneity tests to specify the appropriate unit root, cointegration, and causality tests for the study’s dataset. Thus, CD tests of LM (Lagrange Multiplier) LM CD, and LM_adj. were run, and the results are shown in

Table 5. Results of CD and heterogeneity tests.

Test	Test Statistic	Test	Test Statistic
LM	132 ***	Delta	11.088 ***
LM adj	15.1 ***	Adjusted delta	12.614 ***
LM CD	9.033 ***

*** significant at 1%.

. The H0 hypothesis supporting CD independence is refuted, and the CD availability amongst COEM, RNW, ICT, and ECNDEV is deduced. Further, we used the delta tilde tests, and the results are also introduced in Table 5. The H0 hypothesis supporting homogeneity is denied, and the availability of heterogeneity is deduced. Eventually, the application of causality, cointegration, and unit root tests, sensitive to CD and heterogeneity is required to increase the robustness of the analyses.

The unit roots of COEM, RNW, ICT, and ECNDEV are determined by way of the Pesaran [84] CIPS test, and its results are introduced in

Table 6. Unit root test’s results.

Series	Constant	Constant + Trend
COEM	1.130	0.159
d(COEM)	−4.524 ***	−4.021 ***
RNW	0.511	1.379
d(RNW)	−5.237 ***	−4.272 ***
ICT	0.157	0.478
d(ICT)	−3.671 ***	−4.533 ***
ECNDEV	0.165	0.972
d(ECNDEV)	−4.520 ***	−5.862 ***

*** significant at 1%.

. COEM, RNW, ICT, and ECNDEV all include unit roots with their level values. However, these series seemed stationary when the CIPS test was run with the first-differenced values of COEM, RNW, ICT and ECNDEV.

The long-run connection between COEM, RNW, ICT, and ECNDEV was examined by the LM bootstrap cointegration test, and its results including the LM statistics and probability values are displayed in

Table 7. Results of LM bootstrap cointegration test.

Constant			Constant + Trend
LM Test Statistic	Bootstrap p-Value	Asymptotic p-Value	LM Test Statistic	Bootstrap p-Value	Asymptotic p-Value
1.854	0.870	0.032	7.042	0.980	0.000

Note: Bootstrap probability values are produced from 10,000 simulations, while asymptotic probability values are taken from normal distribution.

. The determination of a cointegration relation between COEM, RNW, ICT, and ECNDEV is grounded on bootstrap p values owing to the CD existence among these variables. Thus, H0 hypothesis of significant cointegration amongst COEM, RNW, ICT, and ECNDEV is approved for both the constant and constant + trend, because the bootstrap p-values are greater than 10%. Consequently, there exists a significant long-run relationship between COEM, RNW, ICT, and ECNDEV. Furthermore, utilization of a cointegration test sensitive to CD prevents us from revealing insignificant cointegration amongst the series by accident.

The long-term effect of RNW, ICT, and ECNDEV on COEM use is forecast by way of AMG estimator, and the coefficients are shown in

Table 8. Cointegration coefficients by AMG estimator.

Countries	RNW	ICT	ECNDEV
Bulgaria	−0.205 ***	−0.061	0.345
Croatia	−0.693 ***	−0.445 **	1.216 **
Czechia	−0.318 ***	−0.221 ***	0.281
Estonia	−0.563 *	−0.610 ***	0.578 **
Hungary	−0.263 ***	−0.419 **	0.513 *
Latvia	−0.297 **	−0.055	0.937 **
Lithuania	−0.206 *	0.085	1.318
Poland	−0.067 **	−0.057	−0.424
Romania	−0.515 ***	−0.031	−0.072
Slovakia	−0.045 **	0.012	−1.664 ***
Slovenia	−0.414 ***	0.229	−2.182 *
Panel	−0.314 ***	−0.295	0.438

***, **, and * show the levels of significance at 1%, 5%, and 10%, respectively.

. The results point out that RNW has a negative impact on COEM at the panel level in all EU transition economies over the long term. On the other hand, ICT seems to negatively impact COEM in Croatia, Czechia, Estonia, and Hungary. Lastly, the long-term impact of ECNDEV on COEM is positive in Croatia, Estonia, Hungary, and Latvia and negative in Slovakia and Slovenia.

The causal connection between RNW and COEM is questioned by way of the E-K causality test, and its results are introduced in

Table 9. Results of causal test between RNW and COEM.

Countries	RNW ↛ COEM		COEM ↛ RNW
	Test Statistics	p Value	Test Statistics	p Value
Bulgaria	0.190	0.663	0.205	0.651
Croatia	9.994	0.019	29.785	0.000
Czechia	3.779	0.286	2.046	0.563
Estonia	0.174	0.676	1.918	0.166
Hungary	1.941	0.164	0.111	0.739
Latvia	5.085	0.024	0.161	0.688
Lithuania	1.713	0.191	0.000	0.982
Poland	0.648	0.421	6.113	0.013
Romania	8.234	0.041	41.075	0.000
Slovakia	0.820	0.365	0.167	0.683
Slovenia	3.584	0.310	1.664	0.645
Panel	38.915	0.014	81.793	0.000

and Figure 1. The panel-level findings demonstrate a bidirectional causal nexus between RNW and COEM. Nevertheless, the findings of the country-level causality test reveal a bidirectional causal relation between RNW and COEM in Croatia and Romania, a unilateral causal relation from RNW to COEM in Latvia, and a unilateral causal relation from COEM to RNW in Poland.

The causal interplay between ICT and COEM is questioned by way of the E-K causality test, and its results are shown in

Table 10. Results of causal test between ICT and COEM.

Countries	ICT ↛ COEM		COEM ↛ ICT
	Test Statistics	p Value	Test Statistics	p Value
Bulgaria	2.555	0.768	115.358	0.000
Croatia	3.261	0.660	23.552	0.000
Czechia	2.251	0.522	2.865	0.413
Estonia	2.613	0.455	16.157	0.001
Hungary	0.250	0.617	1.209	0.271
Latvia	0.951	0.330	0.009	0.923
Lithuania	5.306	0.021	0.034	0.853
Poland	3.919	0.561	4.116	0.533
Romania	0.346	0.556	0.010	0.921
Slovakia	1.709	0.425	1.092	0.579
Slovenia	2.341	0.800	17.482	0.004
Panel	19.603	0.608	152.481	0.000

and Figure 2. The panel-level findings demonstrate a unilateral causal nexus from COEM to ICT. However, the findings of the country-level analysis reveal a unilateral causal relation from ICT to COEM in Lithuania and unilateral causality from COEM to ICT in Bulgaria, Croatia, Estonia, and Slovenia.

Lastly, causal relation between ECNDEV and COEM is questioned by way of the E-K causality test, and its results are shown in

Table 11. Results of causal test between ECNDEV and COEM.

Countries	ECNDEV ↛ COEM		COEM ↛ ECNDEV
	Test Statistics	p Value	Test Statistics	p Value
Bulgaria	4.011	0.548	2.120	0.832
Croatia	34.674	0.000	10.271	0.068
Czechia	20.603	0.001	6.297	0.278
Estonia	10.860	0.054	1.440	0.920
Hungary	4.269	0.511	10.071	0.073
Latvia	13.044	0.023	18.257	0.003
Lithuania	4.983	0.418	7.791	0.168
Poland	10.324	0.067	0.869	0.972
Romania	1.516	0.218	1.927	0.165
Slovakia	7.893	0.162	90.512	0.000
Slovenia	2.743	0.098	0.006	0.936
Panel	74.826	0.000	112.506	0.000

and Figure 3. The panel-level findings demonstrate a bidirectional causal nexus between ECNDEV and COEM. Nonetheless, the findings of country-level causality test reveal bidirectional causality between ECNDEV and COEM in Croatia and Latvia, unilateral causality from ECNDEV to COEM in Czechia, Poland, and Slovenia, and unilateral causality from ECNDEV to HUMDEV in Hungary and Slovakia.

5. Discussion

RNW use has much lower harmful emissions and thus makes a crucial contribution to decarbonization in countries. Furthermore, volatility in fossil fuels’ prices, disruptions in fossil fuels’ supply, and concerns related to energy security encourage countries to transition to RNW. All EU transition economies have achieved a significant increase in RNW use with the influence of the EU’s environment and energy policies. The results of the cointegration analysis also indicate that RNW use has a negative effect on CO₂ emissions over the long term, while the outcomes of the causality analysis reveal a significant influence of RNW use on CO₂ emissions. Therefore, our empirical results support the positive environmental benefits of RNW use, compatible with the theoretical expectations. Nearly all of the recent empirical studies have uncovered a negative effect of RNW use on CO₂ emissions [9,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52]. Furthermore, Dogan and Seker [36], Chen et al. [38], Saidi and Omri [40], and Sezgin et al. [47] uncovered a bidirectional causal relation between RNW use and CO₂ emissions, but Inglesi-Lotz and Dogan [37] unveiled a unilateral causality from RNW to CO₂ emissions, while Otim et al. [9] discovered a unilateral causal relation from CO₂ emissions to RNW use. In conclusion, our results along with the related empirical literature confirm that RNW use is one of the most significant instruments to decrease CO₂ emissions.

ICT penetration has remarkably increased across the world in recent years. However, the theories on the environmental effects of ICT usage indicate that ICTs can impact the environment through diverse channels of energy use, energy efficiency, resource management, digitalization, optimization of RNW sources, and economic growth. The results of the causality analysis uncover a significant influence of CO₂ emissions on ICTs at the panel level and in Bulgaria, Croatia, Estonia, and Slovenia and a significant effect of ICTs on CO₂ emissions in Lithuania. Additionally, the results of the cointegration analysis reveal a negative effect of ICTs on CO₂ emissions in Croatia, Czechia, Estonia, and Hungary. In a similar vein, the related empirical literature is also inconclusive. On the one hand, Nguyen et al. [50], Wen et al. [51], Chang et al. [52], Islam et al. [53], and Linghu et al. [54] revealed a negative influence of diverse ICT indicators on CO₂ emissions, while the studies of [55,56,57,58,59] uncovered a positive effect of ICTs on CO₂ emissions. Furthermore, Islam et al. [53], Shabani and Shahnazi [62], and Altinoz et al. [63] discovered a bidirectional causal relation between the two series, while Park et al. [55] revealed a unilateral causal relation from internet usage to CO₂ emissions.

Economic development is one of the factors closely related to CO₂ emissions, because the economic and social activities of individuals can foster CO₂ emissions through energy use, transportation, production, and consumption. In particular, underdeveloped and developing economies prioritize economic development, disregarding its negative environmental effects, but these countries can gravitate towards the development of clean energy and energy-efficient technologies over time, because of the improvements in human development accompanying economic development [22]. Eventually, the interaction between human development and CO₂ emissions may be different dependent on the socio-economic development level of a country in the context of the EKC hypothesis. In this regard, the results of the causality analysis reveal a mutual interplay between human development and CO₂ emissions at the panel level and in Croatia and Latvia, unilateral causality from human development to CO₂ emissions in Czechia, Poland, and Slovenia, and unilateral causality from CO₂ emissions to human development in Hungary and Slovakia. On the other hand, the results of the cointegration analysis uncover a positive effect of human development on CO₂ emissions in Croatia, Estonia, Hungary, and Latvia and a negative effect in Slovakia and Slovenia.

In a similar vein, Akbar et al. [33] discovered a negative influence of the human development index on CO₂ emissions in the OECD members, while Alkan et al. [34] unveiled a negative influence of human development on CO₂ emissions in the OECD, G20, EU, and advanced countries. However, from the limited number of studies running a causality test between two variables, Bedir and Yilmaz [66] revealed unidirectional causality from CO₂ emissions to the human development index in USA, Turkiye, Spain, Poland, Korea, Japan, and Italy, one-way causality from the human development index to CO₂ in Mexico and France, and bidirectional causality in Switzerland and Norway. Sezgin et al. [67] disclosed a bidirectional causal relation between CO₂ emissions and human development in the USA, the UK, Germany, and Japan and unidirectional causality from CO₂ emissions to human development in France, China, Canada, and Brazil. Alola et al. [68] unveiled a causality from PM2.5, PM10, sulfur dioxide, and carbon monoxide to human development and a causal relation from human development to sulfur dioxide and nitrogen oxides. Last, Minh and Ly [69] unveiled insignificant causality between the two variables.

6. Conclusion, Limitations, Policy Implications, and Future Research Directions

Global warming and climate change have become leading global challenges, and CO₂ emissions are one of the significant factors behind global warming and climate change. For this reason, decreasing global CO₂ emissions is vital to making progress in reducing global warming and reaching the most of the Sustainable Development Goals. In this connection, drivers of CO₂ emissions have been widely explored. This study investigates the interplay between RNW use, ICTs, economic development, and CO₂ emissions in the EU transition states through the E-K causality test and LM bootstrap cointegration test.

The limitations of the study are as follows:

The period of the research was limited to between 2000 and 2021, owing to the availability of the ICT index and RNW use.

The ICT index is calculated based on many indicators from different institutions such as the International Telecommunication Union, World Bank, and the UN Statistics Division. Therefore, the measurement errors related to the ICT index were disregarded.

This study concentrated on the effect of RNW use, ICTs, and human development on CO₂ emissions, and other variables affecting CO₂ emissions were disregarded.

The consequences of the causality analysis point out a bidirectional causal relation between RNW use, economic development, and CO₂ emissions and unidirectional causality from CO₂ emissions to ICT at a panel level. However, the consequences of the causality test at a country level indicate that the direction of causality differs between the EU transition economies. On the other hand, the results of cointegration analysis uncover a negative effect of RNW use on CO₂ in all EU transition countries, but the effect of ICTs and human development on CO₂ emissions differs between the countries.

Based on the results of this research, the policy recommendations are as follows:

The results indicate thar RNW use is a significantly effective tool to decrease CO₂ emissions, and this finding confirms the effectiveness of the EU’s current environmental and energy policies. However, the share of RNW use in final energy consumption is about 25% in these countries, and in turn, further RNW investments are required to reach net-zero emissions.

The negative impact of ICTs on CO₂ emissions also supports the related theoretical views. Therefore, educational and institutional policies and financial resources should be designed to support the use of ICTs in improving energy efficiency, resource management, and optimization of RNW sources.

Future studies can be conducted to analyze the effect of ICTs on energy efficiency, resource management, and optimization of RNW sources.