Introduction

Rapid economic growth of countries has necessitated the usage of energy, which is an imperative production factor that affects the welfare of nations (Kartal 2022; Adedoyin et al. 2023; Bekun 2024). Despite its significance as a production factor, energy has detrimental effects on environmental sustainability by reducing air quality. Climate change is a global environmental threat that is mainly caused by CO2 emissions (Kartal 2023; Raihan 2023; Erdogan et al. 2024a; Khan et al. 2024; Li et al. 2024; Tedeschi et al. 2024). The emissions create various problems, such as environmental disasters, and weather extremes causing floods, droughts, and hurricanes (Ayhan et al. 2023; Erdogan et al. 2024b). So, many countries and supranational authorities have decided to take action to mitigate the negative effects of current climate conditions (Dogan et al. 2022; Anser et al. 2023; Kirikkaleli 2023). COP26 Climate Conference has noted the necessity of securing global net zero CO2 emissions and keeping the global warming level at 1.5 degrees, thus ensuring the mitigation of CO2 emissions has become a top precedence for countries (Yuan et al. 2022; Kartal et al. 2023a; Ramzan et al. 2024). One challenge related to CO2 emissions is that despite the motivations of the countries to reach a net carbon zero, CO2 emissions have reached the highest level as of 2021 (IEA 2022a). Considering these developments, the level of CO2 emissions and the means to cut down these emissions have been lying at the heart of various studies.

GBs are one of the means to mitigate CO2 emissions through allocating funds to eco-friendly (e.g., green technology) projects (Sharif et al. 2024). GBs stand as one of the financing tools that promote the firms or projects that serve energy efficiency improvements or improvements related to pollution mitigation or environmental sustainability (Dogan et al. 2022). The main hurdle to net zero is the lack of finance for renewable energy investments. The core of the discussions in COP26 is the need for financing that will create the environment for adjusting to a zero-carbon emitting phase. Most of the participants of COP26 have declared their desire to cut down on CO2 emissions, yet they have highlighted financial insufficiency to switch to renewable energy, which is hindered even more as a result of shifting funds to health expenditures due to the global health crisis called the pandemic. Thus, the downside risks of the pandemic on allocating funds to the gigantic need for renewable energy investments call for a significant increase in GBs that are at the core of financing sources for green technologies.

GBs facilitate green technology investments and other means of eco-friendly innovations, which will ensure the meeting of net-zero emission targets consented to under the accord (Wang et al. 2023). GBs represent a critical mechanism for mobilizing the necessary capital to transition to a climate-resilient and low-carbon economy. Moreover, given the energy security issues that the world has been witnessing due to the Russian-Ukraine war, energy prices have been displaying high volatility that also triggers the necessity of the countries to shift to renewable sources to handle any unexpected energy shocks (Kartal et al. 2023b). The success and performance of GBs are shaped by the broader economic and political context. Contractionary economic policies or economic uncertainties affect investments in GBs. Moreover, investments in GBs depend on energy prices from traditional sources. In other words, lower fossil fuel prices will discourage green energy investments and reduce investment in GBs (Glomsrød and Wei 2018; Lee et al. 2021; Sartzetakis 2020). However, higher fossil fuel energy prices will escalate the viability of green investments, which will increase the demand for GBs.

Energy, which is the most significant production factor, has become even more critical following the tightening of the energy markets and the creation of a global energy crisis due to the pandemic and the Russian-Ukraine conflict (IEA 2022b; Kılıç Depren et al. 2022). Most countries have witnessed a rapid economic recovery in the post-pandemic period and the decline in green investments during the pandemic has not met the energy needs of the world, which has triggered the high reliance on fossil fuels, creating a huge discrepancy between energy demand and supply (IEA 2022c). Russia, being a major global energy exporter (23.6% of the natural gas supply and 12.3% of oil supply) (British Petroleum 2023), deliberately cut down on exports of natural gas and oil to developed countries, which remain opponents in this conflict, has upsurged the prices of energy. These developments alongside the detrimental effects of the pandemic on the transition to green energy (Żuk and Żuk 2022) have magnified the severity of the energy crisis and energy prices (Guan et al. 2023). It should be noted that recent higher prices of non-renewable energy have triggered a boost in adopting renewable energy (Cao et al. 2020; Atems et al. 2023; Sarker et al. 2023). Energy prices have a restricting position in managing the energy services within a country and consequently affect the environmental quality (Fatima et al. 2021). All in all, oil price stands as a sign of changes in environmental effects and the effect of oil on the environment varies (Baloch et al. 2020).

It is noteworthy to highlight that geopolitical risks are still shaping energy markets and energy security while supply issues remain a setback for many economies. Geopolitical risks, wars, invasions, conflicts with energy-supplying nations, and various trade bans have been still affecting global energy markets and energy commodities (Pata et al. 2023a, b, c; Sarker et al. 2023; Zhang et al. 2023). Geopolitical risk has an important effect on crude oil prices as well as renewable energy investments and environmental pollution. Also, the relationship of geopolitical risk with crude oil prices displays very high volatility during political tension times. Geopolitical risks are also reflected in the green energy stocks (Sweidan 2021), which suggests the effects on traditional and renewable energy prices. Boubaker et al. (2022) stress the substantial effects of the developments on energy markets in Ukraine. Geopolitical risks affect the energy markets through three channels. First, geopolitical risks stimulate energy conversion and reduce oil prices, which is propelled by fuel substitution (Rasoulinezhad et al. 2020). Second, conflicts have a negative influence on investor sentiment, which has a further effect on crude oil prices (Ji et al. 2019). Third, conflicts elucidate oil production through disruption of production and demand, which consequently will be reflected in crude oil prices (Noguera-Santaella 2016). Accordingly, these channels affect crude oil prices and shape the supply and demand for fossil fuels, which will affect the tendency to shift the renewable energy eventually leading to different levels of CO2 emissions.

Given the dynamic link between green bonds, energy prices, geopolitical risks, and CO2 emissions, this study aims to probe into this multifaceted relationship to apprehend the complex relationship. CO2 emissions, which are a major contributor to climate change, can be influenced by energy prices and geopolitical risks. Also, the mitigation of CO2 emissions is mostly the primary goal of green projects funded by GBs. In this context, the study handles global CO2 emissions at aggregated and disaggregated levels as the environmental indicators, considers SPGB as the proxy of GBs, COP as the proxy of energy prices, and GPR as the geopolitical risk proxy, applies a novel WLMC approach since it uncovers the dynamic relationship across times and frequencies, and uses high-frequency daily data between 1st January 2020 and 28th April 2023. Thus, this study presents on CO2 emissions that the most influential factor is the geopolitical risk (2020/1–2021/5) followed by green bonds (2021/5–2021/7), energy prices (2021/7–2023/1), and green bonds (2023/1–2023/4); the effects of the variables are weak at lower frequencies, whereas they are strong at higher frequencies; the effects of the influential factors do not differ at sectors; and the effects of the influential factors vary across times and frequencies.

The novelties of this study to the literature are twofold. First, the study considers global CO2 emissions at aggregated and disaggregated levels, which will enable researchers to evaluate whether the effect of each factor varies. Second, the study adopts a novel WLMC method, which facilitates the exploration of the dynamic relationship between the variables by combining wavelet analysis and multiple correlation analysis and stipulates the understanding of the underlying patterns.

The study continues with the following sections: Sect. 2 provides the theory and literature; Sect. 3 introduces the methods; Sect. 4 demonstrates the empirical results; Sect. 5 concludes.

Theoretical background and literature review

In general, GBs provide financing for clean energy projects. Thus, GBs should be associated with decreasing CO2 emissions. In investigating whether GBs have such an effect or not, a variety of studies have focused on country-level or global-level data. Hammoudeh et al. (2020) analyze index-level data to capture the time-varying causal relationship between GBs and other (financial and environmental) variables. The findings note that there is a causality relationship between the allowances of CO2 emissions and GBs. Yet, the counter-argument is not valid that GBs do not have causality to CO2 emissions. Kanamura (2020) questions the greenness of green bonds and The Bloomberg Barclays MSCI and SPGB show positive correlations and rise with WTI and Brent crude oil prices. On the other hand, the Solactive Green Bond Index exhibits negative correlations and declines with COP and Brent crude oil prices.

Meo and Karim (2022) focus on country-level data and consider the effect of GBs on CO2 emissions in the top ten economies with top GB capitalization. The results indicate GBs as the most significant means to reduce emissions yet note that GB market and country-specific market conditions shape the effectiveness of GBs in mitigating CO2 emissions. Marín-Rodríguez et al. (2022) test dynamic co-movements among oil prices, CO2 emissions, and GBs and notice a unidirectional causality from oil price returns to the returns of CO2 futures and a negative dynamic correlation of green bond index on the returns of CO2 futures, with increased correlations during uncertainty periods. In a similar vein, Marín-Rodríguez et al. (2023) study the relationship between oil prices, GBs, and CO2 emissions and point to a strong interdependence between oil prices and returns of CO2 futures at all times and frequencies and similar interlinkage is evident between the green bond index and returns of CO2 futures, while Green Bond Index is the leading one. Moreover, Adebayo and Kartal (2023) delve into the relationship between GBs, oil prices, and the COVID-19 pandemic on industrial CO2 emissions in United States of America and highlight the time and frequency-varying effects of GBs on the emissions.

In the literature, an ample amount of research has investigated the links between crude oil prices and environmental effects. Several studies have noted a direct link between increasing energy prices and emissions. Saboori et al. (2016) focus on Organization of the Petroleum Exporting countries and the results point out that an increase in oil prices enhances environmental conditions. Nwani (2017) investigates Ecuador, which is an oil exporting country, and noted a link between increasing crude oil prices and economic conditions that are associated with higher CO2 emissions. On the other hand, contradictory findings are also evident. Alshehry and Belloumi (2015) consider Saudi Arabia and suggest that increasing prices of oil create increased usage and have detrimental effects on the environment. Same results are determined for Venezuela by noting dynamic linkages between oil prices, economic growth, and CO2 emissions by Agbanike et al. (2019). Ullah et al. (2020) analyze the highest ten carbon emitters and their results display asymmetric findings. Negative shocks diesel prices accompany reductions with CO2 emissions in China and India. Achuo (2022) scrutinizes the crude oil price shocks and environmental quality connection in sub-Saharan African countries and notes a positive long-run relationship between crude oil price and CO2 emissions. A similar increasing effect of crude oil prices on environmental degradation is defined by Ulussever et al. (2023) for Gulf Cooperation Council countries except for Saudi Arabia and Bahrain.

Zaghdoudi (2017) explores the causality between oil prices, renewable energy consumption, CO2 emissions, and economic growth, and the results propose a negative significant relationship between oil prices, renewable energy, and CO2 emissions. Also, the findings underline a bidirectional relationship between carbon dioxide emissions and oil prices both in the short and long-run. Analysis of China by Zaghdoudi (2018) uncovers the non-linear strong effect of oil prices on CO2 emissions, with supporting evidence both in the short and long-run. Besides, the study unveils that oil price increases escalate CO2 emissions more than their decline. Mahmood et al. (2022) unearth a positive and asymmetrical effect of oil prices on emissions in GCC countries. Similar results are determined by Mohamued et al. (2021) for 26 European countries, 22 oil-producing countries, China, and United States of America with varying evidence for oil-exporting and oil-importing countries. The results suggest that oil price increases in oil-importing (oil-exporting) countries decrease (upsurge) CO2 emissions.

There is little contrasting evidence, which suggests that oil prices have a declining effect on CO2 emissions. Hammoudeh et al. (2014) adopt a quantile framework and observe that an increase in crude oil prices creates a significant decline in carbon prices when the crude oil is at very high quantiles. Wang and Li (2016) explore the 2014 period when crude oil prices declined by nearly half. The analysis reveals that cheaper oil prices create an opportunity to remove fossil fuel subsidies and carbon tax can be introduced during such periods. Okwanya et al. (2023) consider African countries and perceive an asymmetric relationship between oil prices and CO2 emissions, where positive (negative) changes in the oil price cause a reduction (increase) increase in CO2 emissions. The degree of relationship is higher for oil-importing countries.

Moreover, geopolitical risks have long been recognized. Consequently, geopolitical risks act as determinants of huge price hikes and descends that have effects on almost all issues. Despite the significance of the geopolitical risks on the values of green investments and environmental effects, limited studies have focused on this relationship. In the literature, some studies consider similar determinants as a proxy of geopolitical risk (e.g., terrorist activities, political instability) and consider their effects on the environment. Danish et al. (2019) describe the adverse effects of governance indicators, which include political stability and government efficiency, on CO2 emissions in Brazil, Russia, India, and China. Sohag et al. (2019) investigate Türkiye and note that militarization is detrimental to green economic growth. Bildirici and Gokmenoglu (2020) focus on eight countries and find that terrorism is linked to increased CO2 emissions. Also, Bildirici (2020) links terrorist activities with higher CO2 emissions in China, India, Israel, and Türkiye. Wang et al. (2022) investigate the connection between geopolitical risk index and CO2 emissions and it is seen that economic activities are altered through geopolitical risk factors (e.g., trade disputes, military organizations, and energy issues), which in turn affect CO2 emissions. Further, that study highlights that increasing geopolitical risks can lead to higher energy consumption and more military-related activities, potentially hampering research, and development in renewable energy. This limitation in innovation may result in the escalation of CO2 emissions. Tang et al. (2023) investigate the asymmetric effects of various geopolitical risk indicators (e.g., geopolitical acts indices) on crude oil and GB returns. The results show that GB returns are negatively affected by such geopolitical risk indicators in the long-run. Fluctuations in geopolitical risks can also have positive effects by promoting energy independence and encouraging investment in green energy and advanced technology projects (Sweidan 2021). These initiatives can contribute to reducing CO2 emissions.

Although the literature includes various studies, there are some still lack points that no study considers the global case in uncovering the dynamic relationship between GBs, energy prices, GPR, and CO2 emissions at the global level by considering the sectoral differences as well as time and frequency -based varying structure. Accordingly, the study makes a sectoral analysis for the global by applying a novel WLMC approach to close the gap.

Methods

Data

This study investigates the dynamic effects between SPGB, COP, GPR, and CO2 emissions at the global level by making a sector-based analysis. Instead of other environmental indicators that have relatively old and low-frequency data, the present investigation uses CO2 emissions as the environmental indicators to ensure the inclusion of the latest information in line with the studies of Kartal et al. (2022), Adebayo et al. (2023), Hunjra et al. (2023), and Sharif et al. (2023). Also, this study adopts a daily timeframe spanning from 1st January 2020 to 28th April 2023. The selection of this period allows researchers to use up-to-date in the analyses and consideration of recent developments.

Table 1 provides a comprehensive overview of the variables, encompassing their respective symbols, definitions, units of measurement, and sources of data.

Table 1 Variables
Full size table

Estimation models

To thoroughly uncover the dynamic relationships between the variables, a series of models are constructed. They seek to represent the complex and dynamic relationship between the variables. In light of this objective, a total of four different models are developed, which are presented in Table 2.

Table 2 Models
Full size table

These models have been designed to uncover and analyze the complicated connections and dependencies among the variables, shedding light on their mutual effects. By employing these models, a comprehensive and in-depth understanding of the relationships between the variables can be achieved, which provides valuable insights into the underlying mechanisms and dynamics.

Empirical approach

Figure 1 depicts the underlying empirical methodology employed in the analysis. This proposed methodology encompasses six fundamental steps,

Fig. 1
figure 1

Empirical Methodology

Full size image

The methodology begins with examining the preliminary statistics This analysis serves as a crucial groundwork for gaining a comprehensive understanding of the data and variables under investigation. To achieve this, descriptive statistics are examined. Correlation analysis is also performed, enabling the exploration of relationships and dependencies between the variables. As the methodology progresses, the third step is to test the assumption of stationarity (Dickey and Fuller 1979). Stationarity is crucial, as violating this assumption can lead to biased and unreliable results. In the fourth step, a nonlinearity test is conducted to evaluate the linear relationship between the variables, which helps to determine the appropriateness of linear models in the subsequent analysis (Broock et al. 1996). Finally, cointegration analysis is performed to explore the long-term relationship between variables (Pesaran et al. 2001; Kripfganz and Schneider 2020). In the sixth step, WLMC analysis is carried out to examine the coherence or degree of similarity between the variables at different time scales or frequencies (Polanco- Martínez et al. 2020).

WLMC approach

WLMC analysis is a statistical approach that combines wavelet analysis and multiple correlation analysis to explore the relationships between variables in multivariate data sets. It allows for the investigation of relationships at different times or frequencies by providing a comprehensive understanding of the underlying patterns in the data (Polanco-Martínez et al. 2020).

The WLMC analysis begins by applying the wavelet transform to decompose the signals into different frequency bands or scales. The wavelet transform provides a time–frequency representation of the data, revealing the frequency components present in the signals. By decomposing the data, the WLMC analysis can capture both local and global features of the relationships between variables. The results of the WLMC analysis can be visualized to aid in interpretation and understanding. Heatmaps or other visual representations can be used to display the correlation coefficients obtained from the analysis, allowing researchers to identify patterns and trends across different frequencies or scales (Fernández-Macho 2018).

In the WLMC graphs, the first observation corresponds to January 1, 2020. As the data collection progresses, subsequent observations are recorded at regular intervals, such as the 200th observation on October 6, 2020, the 400th observation on July 13, 2021, the 600th observation on April 19, 2022, and the 800th observation on January 24, 2023. The last observation is dated April 28, 2023.

The frequency intervals and their associated frequency terms play a crucial role in the analysis of WLMC figures. These terms reflect the different temporal scales or frequencies at which relationships between variables are explored. The frequency interval of 2–8 represents the short-term scale. It captures relatively rapid fluctuations and transient patterns in the data. Moving to the interval of 8–32, it enters the medium-term scale. The interval of 32–64 corresponds to the long-term scale. Lastly, the frequency term "very long-term" is attributed to the interval of 64–128. Variables analyzed within this range exhibit the most extended timescales in the dataset.

Results

Preliminary analysis

Descriptive statistics

The summary statistics offer an overview of the variables included. The statistics are given in Table 3, which provides key information regarding variables.

Table 3 Descriptive Statistics
Full size table

Among all variables, SPGB has the highest mean value followed by GPR, CO2, COP, INDUSTRY, POWER, and TRANSPORT, in order. Also, GPR has the highest variations followed by COP, SPGB, CO2, POWER, INDUSTRY, and TRANSPORT, respectively. Moreover, based on the Jarque–Bera statistics, it can be stated that all variables, except for CO2 and COP, have a normal distribution.

Correlation matrix

The correlation matrix is presented in Table 4.

Table 4 Correlation Matrix
Full size table

According to Table 4, there are mainly positive correlations between the aggregated and disaggregated level CO2 emissions and the selected variables except for some cases. Specifically, there are negative correlations between CO2 and SPGB; INDUSTRY & SPGB; TRANSPORT & SPGB. So, the correlation results show the changing correlation between the variables.

Stationarity test

The third step of the methodological process entails conducting a stationarity test for each variable. The ADF test is employed to assess the stationarity characteristics of the variables. The results of the ADF test, which serve as indicators of stationarity, are presented in Table 5. This table provides valuable insights into the stationarity properties of the variables, enabling a thorough examination of their time series behavior.

Table 5 Stationarity Results
Full size table

The ADF test results provide insights into the stationarity properties of the variables. The ADF p-values for each variable indicate the presence of unit roots. Based on the ADF test, all variables exhibit statistically significant p-values at the I(1) level. This indicates that the null hypothesis of a unit root (non-stationarity) is rejected in favor of the alternative hypothesis of stationarity for all variables. Furthermore, the decision column indicates that all variables are classified as I(1), indicating that they are integrated into order 1.

Nonlinearity test

The BDS test is employed to examine the nonlinearity of a time series. It evaluates whether the observed series follows a linear pattern or exhibits non-linear behavior. The test compares the autocorrelation structure of the observed series with that of a randomly shuffled version of the series. The results are reported in Table 6.

Table 6 Nonlinearity Results
Full size table

The BDS test results indicate that all variables exhibit statistically significant p-values across different dimensions. This suggests strong evidence of nonlinearity in the series for all variables. Based on these results, the decision column indicates that all variables are classified as non-linear. This implies that the observed series for each variable does not follow a linear pattern and exhibits non-linear characteristics. Therefore, non-linear modeling techniques should be applied as a more suitable approach for analyzing and interpreting the data.

Cointegration test

Table 7 presents the results of the PSS and KS bounds test for the models constructed. The table provides the test statistics (K) and critical values at different significance levels (10%, 5%, and 1%) as well as the corresponding p-values. The p-values indicate the significance of the test results. A smaller p-value suggests stronger evidence against the null hypothesis of no bounds.

Table 7 PSS & KS Bounds Test Results
Full size table

The findings indicate that all models mainly exceed the corresponding critical values, leading to the rejection of the null hypothesis of no bounds.

WLMC results

In the context of the WLMC analysis, the examination of the multiple correlation structure between variables entails the testing of four different models. Model 1 aims to elucidate the relationship between the dependent variable (TRANSPORT) and the explanatory variables (SPGB, COP, and GPR). Likewise, Models 2, 3, and 4 look into the relationships between the dependent variables (INDUSTRY, POWER, and CO2), respectively, and the aforementioned set of explanatory variables (i.e., SPGB, COP, and GPR). The results are presented in Fig. 2.

Fig. 2
figure 2

WLMC Results for Models

Full size image

It can be inferred that the correlations observed between the variables demonstrate statistical significance across all models, mostly within the long-term to very long-term periods. A more detailed analysis reveals that in Model 1, the correlation between variables remains consistently significant at a high level throughout all periods. However, it is noteworthy that the strength of the correlation diminishes from the very long-term to the short-term across all observed periods. Specifically, there is not a statistically significant correlation before October 6, 2020 in the short-term as well as during the period spanning from July 13, 2021, to April 19, 2022. These findings show the temporal dynamics of the correlations between variables within Model 1, highlighting the robustness of the correlations in the long-term, their gradual decrease towards the short-term, and the occurrences, where the statistical significance of the correlations is temporarily diminished.

Consistent with the findings observed in Model 1, the analysis conducted in Model 2 indicates a gradual decline of correlations from the very long-term to the short-term. However, an intriguing disparity emerges in the middle-term, where the correlations exhibit a moderate level of strength, rather than being weak as observed in Model 1. This moderate correlation is evident across almost all periods within the middle-term. Furthermore, it is noteworthy that the correlation coefficients in Model 2 demonstrate a relatively high level of association before October 6, 2020, and after April 19, 2022. This implies that the variables examined in Model 2 exhibit a robust and statistically significant relationship during these specific periods. However, it is important to emphasize that the middle-term is characterized by correlations of moderate strength, which distinguishes it from the weakening trend seen in the short-term.

In both Models 3 and 4, the findings align with those observed in Models 1 and 2, revealing a significant and robust correlation between variables in the long and very long-term, except for the period between July 13, 2021, and August 19, 2021. This signifies a sustained and meaningful relationship between the variables over extended periods, with statistical significance maintained in most instances. Within the medium-term analysis, the correlation between variables remains relatively high during two specific periods: before October 6, 2020, and after April 19, 2022. This indicates the presence of notable associations during these time intervals. However, correlations between variables are generally weak or insignificant in the short-term. This implies a diminished strength of the relationship in the shorter timeframes under examination.

The WLMC heat map analysis reveals valuable insights into the dynamic relationship between the variables across different temporal scales. In the long and very long-term, before July 13, 2021, the GPR variable emerges as the most influential factor affecting CO2 levels. This signifies that GPR plays a prominent role in driving CO2 dynamics during this timeframe. However, there is a notable shift in the most influential factor after July 13, 2021. In the very long-term, the SPGB variable becomes the primary influencer of CO2 dynamics, indicating its heightened significance in shaping CO2 patterns during this period. Conversely, in the smooth term (shorter timeframes), COP takes precedence as the most influential factor, suggesting its dominant role in driving CO2 fluctuations from July 13, 2021, onwards. Furthermore, in the middle-term, it is observed that different factors exert influence on CO2 dynamics depending on specific time ranges. Before October 6, 2020, SPGB is identified as an important factor affecting CO2, highlighting its significance in shaping CO2 patterns during this timeframe. After August 19, 2022, the COP variable assumes a prominent role, indicating its increased influence on CO2 dynamics in the middle-term.

Furthermore, the heat map analysis does not differentiate the CO2 emissions based on sectors. Instead, it provides a comprehensive perspective of the factors affecting CO2 levels across the examined temporal scales without sector-specific differentiation.

Conclusion, policy options, and future research

Conclusion

Degradation in environmental quality has become a sensitive issue for all countries as well as the world. Accordingly, efforts to slow down the negative effects have been increasing. On one hand, countries and international institutions have been putting forth various initiatives and scholars have been searching for the possible causes and potential solutions to this adverse situation on the other hand. While a variety of well-known factors have been considered in the literature, recent knowledge has focused on new factors, such as GBs, energy prices, and GPR. That is why these new factors have been affecting the environmental quality from various perspectives under the current energy crisis and geopolitically risky environment. Also, some countries have a leading role in combating environmental problems, whereas others have a follower role.

Accordingly, this study presents insights about the global case rather than focusing on only one or some countries because “the world is our home”. Hence, the study uncovers dynamic effects between green bonds, energy prices, geopolitical risk, and CO2 emissions by using high-frequency daily data between 1st January 2020 and 28th April 2023 and applying a novel WLMC approach.

It can be summarized that the most influential factor on CO2 emissions is geopolitical risk (2020/1–2021/5) followed by green bonds (2021/5–2021/7), energy prices (2021/7–2023/1), and green bonds (2023/1–2023/4). Also, the effects are weak at lower frequencies, whereas they are strong at higher frequencies. Besides, the effects of the influential factors do not differ according to either aggregated level or disaggregated sectors. Moreover, the effect of the influential factors on CO2 emissions change based on times and frequencies.

The findings of this study are mainly in the same direction as the results of the literature, for instance, Hammoudeh et al. (2020) and Adebayo and Kartal (2023) for GBs; Achuo (2022) and Mahmood et al. (2022) for crude oil price; and Ulussever et al. (2023) for GPR. In addition to presenting consistent results with the current literature, this study further provides new time and frequency-based insights about the dynamic effects between green bonds, energy prices, geopolitical risk, and CO2 emissions by benefitting from the novel WLMC approach.

Policy options

Constructed on the novel WLMC results, various policy alternatives can be argued for the globe. The dynamic relationship between the variables change based on times and frequencies. Also, the effects of green bonds, energy prices, and geopolitical risk on CO2 emissions are stronger at higher frequencies, whereas they are weak at lower frequencies. Hence, global policymakers should care about this fact and adjust their approaches accordingly against various environmental problems. That is why such a time and frequency-based changing effect requires more frequent adjustments in policy framework to obtain a certain success in combating environmental problems.

Besides, the most influential factor change according to periods. Specifically, geopolitical risk is the leading factor for the period between 2020/1 and 2021/5 followed by green bonds (2021/5–2021/7), energy prices (2021/7–2023/1), and green bonds (2023/1–2023/4). Based on this determination, policymakers should think about the most effective factor on environmental quality at the global scale and deal with the most influential factor first, and then go on dealing with other factors as well.

Moreover, the effects of the influential factors on CO2 emissions do not differ according to either aggregated level or disaggregated sectors. In other words, under the different empirical models used, the results are the same, which are aggregated level or disaggregated level effects are similar. For this reason, global policy options can be applied to sub-sectors (transport, industry, and power) as well.

Furthermore, it is critical to understand that green bonds, energy prices, and geopolitical risk do not have a curbing on global CO2 emissions. Although such factors may have a curbing effect in some countries and regions, unfortunately, they are not effective in declining global CO2 emissions. So, global policymakers should focus much more on these factors to turn them into a curbing effect while dealing with other continuous agendas, such as green transition, renewable and nuclear energy stimulates, limiting coal usage (Zeng et al. 2023; Alam et al. 2024).

Future research

The global case in CO2 emissions is examined in this study by some recently critical variables (e.g., GBs, energy prices, and GPR) as the explanatory variables for the period between 1st January 2020 and 28th April 2023 by applying a novel WLMC approach. Based on this reality, new studies can consider analyzing by including the global case and leading CO2-emitting countries at the same time. Also, similar to the approach of this study, new studies can use both aggregated and disaggregated level data simultaneously for further investigation.

Also, because the study uses CO2 emissions as proxy for the environment, new studies can prefer to work with other environmental indicators. Hence other perspectives on environmental quality can be searched in new studies.

Finally, due to the fact that the novel WLMC approach is used in the study, new research can consider applying other recent econometric approaches (e.g., quantile-based as well as wavelet-based other approaches) for further empirical analysis.