The concept of GG denotes the sustainable expansion of the economy in terms of environmental sustainability (Sandberg et al., 2019) and the efficient utilization of natural resources (Reilly, 2012). GG signifies a transformative approach to economic advancement that brings numerous benefits to both communities and ecosystems. One of its primary advantages lies in advancing environmental sustainability, promoting resource optimization, reduced emissions, and biodiversity preservation (D’Alessandro et al., 2020).

The overuse of fossil fuels and the energy needs resulting from industrial activity have caused considerable carbon dioxide (CO2) emissions into the atmosphere. Global warming and the ensuing rise in sea levels are mostly caused by these emissions (Keeling, 1973; Wuebbles and Jain, 2001; Quadrelli and Peterson, 2007; Heede, 2014; Li and Haneklaus, 2021). Recognizing the serious repercussions of greenhouse gases, global initiatives are underway to mitigate carbon emissions. It's increasingly vital for countries to explore avenues for curbing environmental degradation through sustainable development that benefits all (Li, Dong and Dong, 2022; Huang and Zhao, 2022).

It is revealed that positive changes in raw material efficiency reduce greenhouse gas emissions, whereas negative changes lead to an increase, particularly in the long term. Raw material efficiency for GG and environmentally relevant technologies decreases greenhouse gas emissions. The importance of raw material efficiency for GG is highlighted in studies conducted by Wang et al. (2020) and Gyamfi et al. (2021) for the G7 economies, Tufail et al. (2021) for the OECD countries, Adebayo et al. (2022) for newly industrialized countries, Shen et al. (2021) for China, and Bekun et al. (2019) for 16 EU countries. Bleischwitz (2010), for example, shows that although economic expansion raises greenhouse gas emissions over the long run, GG and raw material efficiency for technology connected to the environment reduce these emissions.

Moreover, a strong linkage exists between carbon emissions and innovation within green technology, energy, and environmental technology sectors. Conversely, proponents also posit that green technologies play a pivotal role in mitigating global pollution and advancing towards carbon neutrality (Ulucak, 2020; Dong et al., 2022).

The measurement methods for GG can be broadly classified into two main approaches. One such approach is the Malmquist-Luenberger (ML) index, pioneered by Caves et al. (1982). This index evaluates total factor productivity by employing input or output-oriented directional distance functions. Researchers like Zhang et al. (2024), Fang et al. (2021), and Ngo, T. Q. (2024) have utilized ML-based methodologies to derive green growth indices. Secondly, some methods integrate green growth by quantifying all environmental and resource costs and benefits, as highlighted by scholars such as Huang and Zhao (2022), Xu (2022), and Guo et al. (2017). Additionally, many studies in the literature focus on developing green indicator systems and parameters to gauge green growth effectiveness, including works by Sohag et al. (2019) and Song et al. (2019). Lastly, indices are formulated to assign weightings to specific indicators of green growth, as seen in the research by Kumar (2017), Maji (2019), and Zhao J. et al. (2022).

Green growth (GG), characterized by sustainable economic development that minimizes environmental degradation, is influenced by various factors ranging from trade openness and financial development to environmental regulations, technological innovations, natural resource use, and green technology. This literature review synthesizes key findings from recent studies to shed light on the diverse drivers of GG and their implications.

The convergence and divergence dynamics within sustainability and green economy have been intensely scrutinised in recent academic literature. Four key studies shed light on various aspects of this discourse. Shen et al. (2022) examine how green productivity is calculated, emphasizing the role of technological progress in driving GG in Africa. Eleftheriou et al. (2024) focus on the Sustainable Development Index (SDI), revealing distinct convergence clubs among countries with implications for ecological wellbeing. Sueyoshi and Wang (2020) explore convergence clubs in renewable energy innovations, highlighting factors such as income, emissions, and institutional quality. Finally, Pinar (2024) examines global convergence patterns in renewable energy innovations, underscoring the importance of environmental regulations and R&D investment. These studies collectively contribute to understanding convergence phenomena within the sustainability realm. While existing studies have extensively investigated convergence in various environmental factors (González-Álvarez et al., 2020; Zhang et al., 2021; Ulucak, 2021; Bigerna et al., 2021; Saba and Ngepah, 2022), research on convergence in production-based CO2 productivity remains insufficient. Additionally, despite the examination of convergence in GG, there is a significant vacuum in the literature regarding the drivers of GG convergence.

Shen et al. (2022) employed the Luenberger productivity measurement to incorporate carbon emissions into the measurement of green productivity. The study observed that the annual average green productivity growth in Africa is 1.51%. GG in Africa primarily stemmed from technological progress rather than efficiency changes, with three different convergence clubs identified for technological progress. Eleftheriou et al. (2024) utilized the Phillips and Sul methodology to examine convergence/divergence models of the Sustainable Development Index (SDI) across 137 countries. Their findings identified two convergence clubs: one comprising developing countries from Africa and Asia, and the other consisting of developed countries like the USA, Canada, and Australia. They concluded that the SDI encourages ecological wellbeing, with top-ranking countries characterized by minimal ecological impact and high human development.

Sueyoshi and Wang (2020) conducted a two-stage analysis using Data Envelopment Analysis (DEA) to construct sustainable development indices. They applied the Phillips and Sul algorithm for club convergence analysis and highlighted two convergence clubs in renewable energy innovations: one comprising more innovative countries and the other containing less innovative ones. Their conclusion suggested that countries with higher per capita income, lower CO2 emissions, higher R&D investment, better environmental regulations, and stronger institutional environments have a higher probability of being part of the innovative club. Pinar (2024) utilized the Phillips and Sul convergence algorithm to evaluate the convergence of renewable energy innovations across 90 countries. Determinants were explored through Probit and Logit regression analyses. Pinar (2024) revealed the absence of global convergence in renewable energy innovations, instead identifying two convergence clubs based on innovativeness. The study suggested that countries can enhance their likelihood of joining a more innovative club by increasing investments in R&D, improving environmental regulations, and enhancing institutional quality.

The literature is replete with unremitting debate on the application of convergence of renewable energy, renewable energy innovations, green productivity, SDI, and green economic efficiency among others. However, there is limited literature on the convergence of green economic growth and the drivers. This study contributes to the literature by (1) examining the global convergence path of green economic growth and (2) investigating the drivers of this convergence. This analysis is extended to regional analysis to understand the regional dynamics.

This study encompasses a panel of 134 countries spanning the years 1995–2022. The selection of countries was influenced by data availability, detailed in Supplementary Appendix Table A1 in the Appendix. Green economic growth, which is the dependent variable, is quantified using resources and environmental productivity. Specifically, we utilised production-based CO2 productivity (US dollars per kilogramme, 2015). This approach aligns with previous studies by Kasztelan (2017), Porada Rochoń (2021), Wu et al. (2022), and Kirikkaleli and Addai (2023). In exploring the drivers of club convergence, we incorporate economic growth, measured by per capita real GDP (in constant 2015 US dollars), drawing on the empirical insights of Fernandes et al. (2021) suggesting its contribution to green economy convergence. Additionally, trade openness is introduced, measured by the addition of imports and exports divided by gross domestic product is consistent with the approach of Shahbaz et al. (2014).

Given the empirical outcomes of Tariq et al. (2024) indicating the influence of environmental pollution (CO₂ emissions per capita) on green economic growth, we include CO₂ emissions in the model. Population size (total population) is integrated into the model based on the proposition of Nathaniel et al. (2021), highlighting its association with increased energy demand and environmental impacts. In order to evaluate the utilisation of renewable energy, we use renewable energy demand as a proportion of the overall final consumption of energy. This approach aligns with the findings of Ashfaq et al. (2024) concerning the critical role that the transition to renewable energy plays in fostering green economic growth. We used institutional quality measured by the average of control of corruption, regulatory quality, political stability, rule of law, government effectiveness, and voice and accountability. This metrics has been widely referenced by (Konara and Shirodkar, 2018; Elamer et al., 2020; Tunyi et al., 2020), providing insights into its impact on the green economy.

Table 2 presents the results of the PS log(t) test conducted on a global scale, encompassing data from 134 countries. The table comprises two panels: A showcases the log(t) coefficients and associated t-statistics for the entire sample, both the original and merged club classifications, while Panel B provides the log(t) values specifically for the final club classification. The findings from the log(t) convergence test indicate an absence of convergence in green economic growth across the sampled countries. This conclusion is drawn from the coefficient of log(t) (−0.0936) and its associated t-statistics (−7.8996), which fall substantially below the critical threshold of −1.65. This divergence in green economic growth trends among the 134 countries suggests significant disparities. These results align with prior research highlighting variations in the green economy and environmental quality across nations (Porada Rochń, 2021; Haider et al., 2021).

Furthermore, PS underscores the opportunity to explore club convergence alongside analyzing convergence within the entire panel. Initially, the club clustering algorithm identified three convergence clubs across the full panel. Following PS’s guidance, a club convergence algorithm was subsequently deployed to mitigate the risk of overestimation. The results of the club merger tests indicate the potential combination of Club 1 + 2 and Club 2 + 3. Remarkably, the t-statistics for each club (Club 1: 8.458; Club 2: 1.813) surpasses the critical threshold of −1.65. Consequently, the empirical discoveries stemming from the club clustering algorithm affirm that the club convergence null hypothesis is accepted for both clubs.

Table 3 provides a comprehensive list of countries constituting each club formation. Furthermore, the relative transition curves for Clubs one and two are shown in Figures 1, 2, respectively, while Figure 3 illustrates the average transition paths for both clubs. Upon scrutinizing these curves, it becomes evident that the transition pathways for each club adhere to the theoretical framework of club convergence. According to this framework, members of each club have relative transition paths that converge towards different constants. A meticulous visual analysis of these data points suggests that while the green economy may vary among countries within each club, there is a notable tendency towards substantial convergence.

Converging of GG globally is crucial for addressing urgent environmental challenges such as deforestation, pollution, biodiversity loss, and climate change. This is the main objective of the Environmental Sustainability Goals (SDG) set forth by the United Nations in 2019. By transitioning to sustainable practices and reducing resource consumption, countries can collectively mitigate environmental degradation and preserve the Earth’s natural ecosystems for future generations. Since green economy convergence emphasizes the decoupling of economic growth from environmental degradation, its divergence or convergence implications are crucial.

The study’s findings about the global divergence of the green economy imply that diverse causes exert pressure on the green economies of different countries around the globe. Thus, universal laws pertaining to environmental sustainability might not work everywhere in the world. Because different countries have different starting conditions and different pathways towards the long-run equilibrium, it may be challenging to formulate common environmental policies. Hence, it is imperative to align international environmental accords and policies according to these club classifications (Pettersson et al., 2014; Stern, 2017). Furthermore, comparable strategies are unlikely to be successful in nations where associated variables differ, as stated by Ahmed et al. (2019). In addition, considering the country-specific drivers of the green economy implies the possibility of implementing tailored environmental policies at the national level, in addition to the common club-classification measures.

Referring to Table 4, the convergence log(t) test results for each global region are delineated. The table is partitioned into two panels: Panel A presents the initial club classification, merged clubs, as well as the coefficients and associated t-statistics of log(t) for the entire sample within each region. Panel B delineates the final club classifications. Our analysis encompasses four continents: Africa, Asia, Europe, and Latin America, which also includes North America and the Caribbean. The respective t-statistics of −40.3704, −1.9909, and −10.7441, all markedly falling below −1.65, suggest a lack of evidence supporting conditional or relative convergence across the entire samples for Africa, Europe, Latin America, and the Caribbean. Consequently, the null hypothesis of full panel convergence is rejected for each region.

However, as the t-statistic of 4.4805 is greater than −1.65, the estimates of the log(t) convergence tests for Asia show evidence of conditional/relative convergence. As a result, the null hypothesis of full panel convergence is accepted. Furthermore, if the estimate of log(t) is larger than or equal to 2, it is also possible to determine the size of the corresponding log(t) coefficients, which, in accordance with PS, represent convergence patterns or decay rates. The full sample of the aforementioned regions (Africa: −0.6764; Asia: 0.2084; Europe: 0.0284; Latin America, North America, and Caribbean: −0.2503) has a coefficient of less than 2, as indicated in Table 4, indicating that there is a divergence in green economic growth for each of the four regions.

An examination of possible convergence club was subsequently done. The algorithm uncovers four convergence clubs for Africa at first, but after using a club merging technique to prevent overestimating the clubs, the final club classification was reduced to three, as shown in Panel B of Table 4. Club 1 comprises two countries, while Club 2 and Club 3 consist of twenty and eleven countries, respectively. Table 5 presents the list of countries included in each final club classification for Africa. With decay rates of 36.9%, 17.8%, and 13% for clubs 1, 2, and 3, respectively. Table 4 indicates that there is conditional convergence among the clubs, with each log(t) coefficient being significant and having a t-test greater than −1.65. Based on their relative log(t) coefficients, club 1 and club 2 appear to have the highest rates of convergence. The empirical results support the findings of Arogundade et al. (2023) and Ulucak et al. (2020), who indicated that three convergence clubs exist in the region despite finding divergence in environmental quality in Africa.

Now turning to the findings of the Asian region. The entire sample for the Asian continent shows indications of convergence, in contrast to the results of the world sample and the African continent. Given that the log(t) t-statistic of 4.4805 is greater than −1.65, the null hypothesis for entire sample convergence is accepted. This demonstrates that, at a decay rate of 20.84%, the green economies of Asian countries do, in fact, converge to the same steady state. The club clustering algorithm identified a single club that is representative of the full sample. For the region, the club merging technique did not need to be used because the final club categorization consisted of just one club.

The club clustering algorithm results in a final club categorization of one for European countries, whereas the club clustering algorithm initially identified three clustering clubs. An examination of the results indicates the existence of convergence (since the t-test for club one is 4.463 > −1.65). The empirical findings corroborate the conclusions drawn by Yıldırım et al. (2021), who observed no disparities in environmental quality across 16 European countries. Furthermore, they align with the discoveries of Ulucak and Apergis (2018), who identified three convergence clubs among EU countries demonstrating convergence in environmental quality.

Lastly, the club clustering’s empirical results for Latin, North America, and the Caribbean indicate a divergence since the log(t) t-statistic of −10.7441 is less than −1.65 (the null hypothesis for entire sample convergence is rejected). This confirms the divergence in green economic growth of the Latin, North America, and Caribbean countries. The club merging technique was implemented to mitigate the risk of overestimating club memberships, resulting in a reduction of the final club classification to two, as illustrated in Panel B of Table 4. Club one comprises 17 countries, whereas Club 2 consists of 7 countries. The list of countries within each final club classification for the region is provided in Table 5. Notably, with decay rates of 28.2% and −0.45% for Clubs 1 and 2, respectively, Club 1 exhibits the highest rate of convergence based on its relative log(t) coefficients.

Following the proposition of Quah (1990) and Carlino and Mills, (1993), Carlino and Mills, (1996) that stationarity tests can be used to assess stochastic convergence. These studies argue that stochastic convergence occurs when a series is trend stationary. On the other hand, the presence of a unit root indicates that the shock is permanent, leading to the divergence of the series from the sample mean. Due to this, our study applied the stationarity test to assess whether there is convergence or divergence in green economic growth among the selected sample. We adopted the Fourier ADF unit root test since the traditional stationarity test (ADF and PP) cannot address structural breaks. As suggested by Enders and Lee (2012), the FADF unit root test exhibits superior power performance as it accounts for structural break and nonlinearity.

As presented in Figure 4, FADF test statistics for 48 countries are statistically significant since the p-values are less than a 5% significance level (For a list of these countries, see Supplementary Appendix Table A1 in the Appendix). Hence, green economic growth for these countries is stationary at a 5% significance level. Put differently, these countries' green economic growth converges to the World average of green economic growth. However, since the p-values of the remaining countries are not significant, we can conclude that there is a divergence in green economic growth. In other words, these countries' green economic growth does not converge to the World average of green economic growth.

Table 6 displays the average marginal effects on club membership probability, derived from both an ordered logit and probit model. It's crucial to acknowledge that while the size of the estimates may not inherently provide a straightforward economic interpretation, assessing the significance of explanatory variables in predicting club membership is vital before delving into economic implications. This evaluation is conducted by calculating the marginal effects of predicted probabilities, clarifying how a one-unit rise in an explanatory variable impacts the probability of membership in each club while keeping all other variables fixed at their respective sample averages (Potoski and Prakash, 2005; Bartkowska and Riedl, 2012; Von Lyncker and Thoennessen, 2017). Comprehension of the sign and magnitude of estimation coefficients is critical, as they indicate whether a variable amplifies or diminishes the likelihood of convergence into a particular club. To streamline our analysis, we primarily rely on empirical estimations from the ordered logit model.

The estimated results from Table 6 unveil intriguing insights. A 1% increase in real GDP correlates with a significant reduction in the probability of converging into Club 1, specifically by 0.842%. This finding suggests that elevated GDP growth often accompanies heightened energy consumption, which, in turn, exacerbates reliance on carbon-intensive fossil fuels such as coal, oil, and natural gas. Such dependence on carbon-centric energy sources may hinder the transition to sustainable practices and renewable energy, consequently impeding efforts to achieve convergence within the green economy.

Conversely, while the impact of real GDP growth on the probability of converging into Club 2 is statistically insignificant, there exists a marginal increase. A 1% rise in real GDP elevates the probability of converging into Club 2 by 0.452%. This finding prompts the speculation that economic growth could potentially foster innovation and technical advancement, leading to the development of novel strategies and methodologies that promote environmental sustainability.

To expedite the convergence of the green economy, nations experiencing faster GDP growth rates might consider allocating greater resources towards research and development initiatives in critical areas such as sustainable agriculture, waste management, renewable energy, and green transportation. Such strategic investments could catalyze transformative progress towards a more sustainable and environmentally resilient future.

Furthermore, the estimated coefficient of the squared term of real GDP indicates a positive and significant relationship for Club 1 and negative and significant relationship for Club 2. This infers that the nature of the relationship between the economy and the club membership of green economic convergence is characterized to be nonlinear (convex for Club one and concave for Club 2).

Trade openness exerts a positive and statistically significant influence on Club 1, while demonstrating a negative and statistically significant impact on Club 2. To be precise, a 1% increase in trade openness enhances the likelihood of converging into Club one by 0.2%, whereas it diminishes the probability of converging into Club two by 0.55%. We believe this is because trade liberalisation makes it easier for countries to share information, best practices, and environmentally beneficial technologies. Similarly, countries can obtain green technologies, expertise, and innovations created elsewhere through international commerce, which speeds up the adoption of sustainable practices and fosters convergence towards green economic objectives.

Furthermore, the estimated coefficient of CO2 emissions reduces the probability of club membership of Club 1 by 0.76% and increases the probability of joining Club 2 by 0.72%, if CO2 emissions increase by 1%. A possible explanation for the negative impact on Club 2 convergence could be because CO2 emissions contribute to air pollution, climate change, and ecosystem disruption in the environment. By reducing natural resources, endangering biodiversity, and jeopardizing ecosystem services necessary for sustainable development, this degradation weakens the basis of a green economy. However, we believe that the positive impact of Club two membership could be because countries may become increasingly aware of the need to address the climate change impact of CO2 emission. This acknowledgment may inspire initiatives and regulations meant to further sustainability and the shift to a green economy.

Population size contributes negatively to the probability of club membership of Club one by 0.3% and 0.03% to club membership of Club 2. The intuition behind this empirical outcome is that the need for natural resources like energy, land, and water rises as the population grows quickly. This increased demand has the potential to undermine efforts towards sustainable resource management and the convergence of the green economy by causing overexploitation, deforestation, habitat damage, and resource depletion. Similarly, natural resources contribute negatively to the probability of Club 1 membership by 0.04%. However, it increases Club 2 membership by 0.12%. For the negative impact of natural resources, we believe that natural resource exploitation, particularly when done in an unsustainable way, can result in habitat destruction, biodiversity loss, and environmental deterioration; all of which compromise the environmental sustainability required for the convergence of the green economy. However, the positive impact of club membership can be because countries with abundant natural resources, including those with geothermal, wind, or solar power, can use these resources to generate renewable energy and promote convergence of the green economy.

The estimate of renewable energy transition is negative and statistically significant, indicating that renewable energy reduces the probability of club membership of Club 1 and Club 1 by 0.003% and 0.12%, respectively. This suggests that the adoption of renewable energy might not necessarily translate to green economic growth convergence among countries. This is because some countries might have potential challenges like financial constraints, transition challenges, and technological limitations towards the adoption of renewable energy transition. Similarly, the coefficient of institutional governance reveals that it decreases the probability of club membership for Club 1 and 2 by 0.21% and 0.1%, respectively.

The study further examines the regional analysis on the drivers of club convergence of green economy. The empirical outcomes suggest that the relationship between economic growth and green economic growth convergence is characterized to be nonlinear. However, the nature of this nonlinearity differs across regions. For instance, it is convex for Africa and Europe, while the relationship is concave for Asia and Latin, North America, and Caribbean. Trade contributes positively to club membership of Club one across all the different regions. However, it negatively impacts the probability of Club two membership across all the regions.

CO2 emissions reduce the probability of club membership of Club one in Latin, North America, and Caribbean, Europe, and Asia. While it increases the club membership of Club one in Africa. Nonetheless, CO2 emissions increase the probability of club membership of Club 2 Latin, North America, and Caribbean, Europe, and Asia. Whereas the impact in Africa is negative. With the exception of Europe, population contributes positively to the probability of Club one membership. However, it contributes negatively to Club two membership in Africa, Asia, Latin, North America, and Caribbean, while it contributes positively to Europe. Furthermore, the estimated coefficient natural resources contribute negatively to club1 membership in Asia, Europe, and Latin, North America, and Caribbean, while it impacts positively in Africa.

Renewable energy transition is negative and statistically significant, suggesting that renewable energy reduces the probability of club membership of Club one in Asia, Europe, and Latin, North America, and Caribbean, while it impacts positively the probability of club membership in Africa. However, renewable energy transition is positive and significant on the probability of club membership of Club two in Asia, Europe, and Latin, North America, and Caribbean, but the impact is negative for countries in Africa. The empirical outcome of institutional governance contributes negatively to the probability of Club one membership in Africa and Latin, North America, and Caribbean, while it has positive and significant impact on probability of Club one membership in Asia and Europe. Nonetheless, institutional governance increase the probability of Club two membership in Africa and Latin, North America, and Caribbean, whereas, it has a significant and negative impact on Club two membership in Asia and Europe.