Environmental sustainability has emerged as a critical challenge for BRICS nations—Brazil, Russia, India, China, and South Africa—as they experience rapid economic growth, urbanization, and digital transformation (Farooq et al., 2023; Zhao et al., 2024). These developmental shifts, while driving industrial progress, have also intensified environmental pressures, particularly in the form of rising carbon emissions and ecological degradation (Caglar et al., 2023; Esmaeil et al., 2020; Kongbuamai et al., 2020; Tariq et al., 2024). Countries like China and India face increasing health and environmental risks linked to air pollution and fossil fuel dependency, highlighting the urgency of aligning economic development with sustainability goals (Dincă et al., 2022; Eyuboglu et al., 2024; Ma et al., 2025; Sajith Kumar et al., 2022).

Ecological sustainability, as used in this study, refers to the capacity of natural systems to maintain their essential functions, diversity, and productivity over time while supporting human development needs (Caglar et al., 2024). To operationalize this concept, we utilize biocapacity as a proxy indicator of environmental sustainability. Biocapacity measures the biological productivity of ecosystems—specifically their ability to regenerate resources and absorb waste, including carbon emissions. It comprises several land-use components: cropland, grazing land, forest land, fishing grounds, and built-up land. These components collectively reflect the ecological pressure exerted by human activities and the regenerative capacity of the environment. Environmental sustainability demands a balance between resource consumption and regeneration, and its attainment requires not just technological innovation but also institutional commitment and societal behavioral change (Buhari et al., 2020; Gao and Fan, 2023). Closely linked is the concept of resilience, which denotes the ability of ecosystems and social systems to absorb disturbances, adapt, and reorganize without compromising long-term functionality or the continued provision of ecosystem services (Gayen et al., 2024; Luo et al., 2024). In the context of BRICS, resilience encompasses the adaptability of environmental policies, urban systems, and human capital to shifting socio-economic and ecological conditions.

The global urgency of these challenges is articulated in the United Nations Sustainable Development Goals (UNSDGs)—a blueprint for achieving a better and more sustainable future by 2030. Several goals are directly pertinent to this study: SDG 4 (Quality Education), SDG 9 (Industry, Innovation, and Infrastructure), SDG 11 (Sustainable Cities and Communities), and SDG 13 (Climate Action). These goals underscore the need to integrate environmental objectives into education, technology, governance, and economic development (Akrofi et al., 2022). Moreover, SDGs 14 (Life Below Water) and 15 (Life on Land) are foundational to the entire SDG framework, acting as multipliers of co-benefits across the goals. Conservation of biodiversity under these goals supports the achievement of other SDGs by enhancing ecosystem services, promoting resilience, and sustaining livelihoods. This interdependence highlights the necessity of incorporating biodiversity considerations into strategies aimed at achieving the broader SDG agenda (Obrecht et al., 2021).

In this context, education (EDU), digital innovation (DI), and urbanization (URB) represent three pivotal forces shaping environmental outcomes in emerging economies. EDU is frequently cited as a driver of sustainable behavior and policy innovation (Li et al., 2023). However, its influence is not monolithic. For this study, we adopt the standard classification where primary and secondary education represent lower educational levels, while tertiary education (such as university or vocational degrees) constitutes higher education. Tertiary education is often linked to ecological responsibility and technological advancement, while lower education levels have been historically associated with industrial expansion and increased environmental strain (Haseeb et al., 2023; Kalaycı Alas and Korutürk, 2024; Li et al., 2025), suggesting a nonlinear relationship between educational attainment and environmental outcomes.

Building on this, digital innovation (DI) also plays a complex role. Emerging technologies such as AI and IoT can facilitate sustainability goals through smarter systems and efficiencies but may simultaneously lead to greater emissions and electronic waste if deployed without robust regulatory oversight (Adebayo et al., 2022; Dilanchiev and Taktakishvili, 2021; Hu et al., 2024; Lisha et al., 2023; Ramzan, 2021). Thus, understanding the governance of digital transformation becomes critical, especially in contexts where institutional capacity may be uneven (Li and Jianxing, 2024; Sinha et al., 2020a; 2020b; Waris et al., 2023).

Urbanization (URB), the third focal parameter, continues to reshape socio-environmental systems. While structured urban development can promote sustainability through efficient infrastructure and land use, unregulated expansion frequently contributes to habitat loss and pollution (Mehmood et al., 2023; Yang et al., 2024). The interplay between URB and environmental policy stringency (EPS) is also crucial, as the real-world impact of such policies depends heavily on their design and enforcement, which vary across national contexts (OECD, 2022).

Given these complexities, this study investigates how EDU, DI, URB, EPS, and economic growth (E.G.,) influence the sustainable environment (SE) in BRICS nations from 1990 to 2023. In doing so, it seeks to fill critical gaps in the literature by:

• Examining the nonlinear effects of education on sustainability,

A key methodological contribution of this study lies in its use of a multi-analytical approach to explore the environmental impacts of education, digital innovation, and urbanization across BRICS nations. Specifically, we employ Method of Moments Quantile Regression (MMQR) to capture variations in these relationships across different levels of sustainability performance—an advancement over conventional mean-based models that often overlook such distributional nuances. Additionally, Feasible Generalized Least Squares (FGLS) is applied to enhance robustness against heteroscedasticity, and Granger causality analysis is used to examine potential directional influences among variables. Together, these methods provide a more differentiated and context-sensitive understanding of sustainability dynamics in emerging economies.

By integrating education, technology, urbanization, and policy into a unified analytical framework, this research advances a comprehensive model of sustainability planning tailored to emerging economies. Its alignment with SDG commitments and COP climate agreements ensures relevance for both national policymakers and international development agendas.

This study formulates the following hypotheses to guide empirical investigation:

H1

Higher levels of educational attainment—particularly tertiary education—will have a positive and significant effect on environmental sustainability across BRICS nations, reflecting the role of education in promoting environmental awareness and policy engagement.

H2

Increased digital innovation is expected to exert a positive influence on sustainability outcomes by enhancing efficiency and supporting green technologies, although its benefits may be offset if not managed with appropriate environmental safeguards.

H3

Urbanization, when accompanied by structured infrastructure and planning, will have a positive and significant impact on environmental sustainability through resource optimization and reduced per capita emissions.

H4

Both economic growth and environmental policy stringency are hypothesized to have a positive and significant association with sustainability performance, indicating the importance of institutional and economic drivers in shaping environmental outcomes.

These hypotheses are framed to reflect testable, directional expectations based on existing literature and are intended to inform empirical analysis and policy discussions on sustainability in emerging economies.

The subsequent sections are detailed as follows. Section 2 highlights the importance of leveraging digital innovation and education in urban settings to foster environmentally sustainable communities. Section 3 elaborates on the empirical methods used to examine the relationships between environmental policies, urbanization, education, urban population, and environmental health. Section 4 presents the key findings and discussions. Finally, the concluding section offers policy recommendations and final reflections.

This section presents a review of the literature exploring the connections between education, ecological policies, urbanization, digital innovations, and environmental sustainability.

The results of the cross-section dependence (CD) test in Table 4 reveal that most variables—EG, URB, EDU, DI, and EPS—exhibit significant cross-sectional dependence, indicating that developments in one BRICS country tend to influence others. This finding aligns with the expectation of strong regional interconnectedness, particularly in areas such as economic growth, technological advancement, and policy responses to global environmental challenges. However, in contrast, SE (proxied by biocapacity) does not exhibit significant cross-sectional dependence (p-value = 0.276), suggesting that biocapacity is shaped largely by country-specific conditions. This apparent isolation may stem from substantial heterogeneity in natural resource endowments, geographic and ecological systems, and the structure of land use and conservation policies across BRICS nations. Unlike policy- or technology-driven variables, biocapacity is deeply rooted in physical and biological characteristics that are not easily transferrable or influenced by external shocks. Moreover, national-level ecological strategies and varying commitments to biodiversity and land preservation may further reinforce this divergence. While this outcome may appear to challenge assumptions of regional environmental interdependence, it underscores the need to differentiate between shared policy challenges and localized ecological capacities. The CD results thus provide important justification for the use of robust estimation techniques like MMQR and FGLS, which can accommodate both interdependent and independent cross-sectional structures.

The slope heterogeneity test results in Table 5 reveal key differences between the two model specifications. For the linear (SE = f (E.G., URB, EDU, DI, EPS)) and nonlinear (SE = f (E.G., URB, EDU, EDU², DI, EPS)) models, both the standard and adjusted test statistics are highly significant at the 1% level, confirming the presence of slope heterogeneity across countries. This suggests that the impact of explanatory variables on SE differs substantially among BRICS nations, reinforcing the need for estimation techniques like MMQR that account for such variation.

To achieve trustworthy and effective estimates, it is essential to eliminate time-dependent elements from the data to assess its stationarity before implementing any robustness or regression analyses (Hu et al., 2022). The unit root diagnostics in Table 6, conducted through CADF and CIPS tests, reveal a mixed integration order among the variables, with critical implications for model specification. E.G., and EDU are found to be stationary at the level under both tests, indicating I (0) behavior, while DI and EPS achieve stationarity only after first differencing, confirming they are I (1). However, SE and URB remain non-stationary even after first differencing in both CADF and CIPS tests, suggesting potential issues with their integration properties. This persistent non-stationarity could reflect structural shifts, measurement inconsistencies, or long-term trends not captured by standard differencing. Therefore, the application of robust estimation methods like MMQR and FGLS is particularly appropriate, as these techniques can accommodate non-stationary panels and provide consistent estimates in the presence of unit root and slope heterogeneity.

The results of Westerlund’s (2007) cointegration test in Table 7 provide mixed but insightful evidence regarding the long-run equilibrium relationships among the variables in both model specifications. For the linear model, the Gt and Pt statistics are highly significant at the 1% level, indicating strong evidence of cointegration based on group-mean and panel-based t-statistics. However, the Ga and Pa statistics are not significant, suggesting that when averaging across units, cointegration is not uniformly present—highlighting potential heterogeneity across BRICS nations.

In the nonlinear model, a similar pattern emerges. The Gt (−7.712) and Pt (−37.513) statistics remain highly significant, reinforcing the presence of cointegration based on t-statistics, even when education is modeled nonlinearly. Yet again, the Ga and Pa statistics are not significant, suggesting that cointegration is not consistent across all cross-sectional units when using average-based approaches.

Overall, these results confirm that long-run relationships exist among the variables in both models, particularly when considering individual panel dynamics (Gt and Pt), but also highlight the importance of accounting for cross-sectional heterogeneity, justifying the use of distribution-sensitive techniques such as MMQR.

The Variance Inflation Factor (VIF) results presented in Table 8 indicate that multicollinearity is not a concern in the model. All VIF values fall well below the conventional threshold of 10, with the highest being 2.950 for EDU and a mean VIF of 2.180. The corresponding 1/VIF values, ranging from 0.339 to 0.741, further confirm acceptable levels of correlation among the explanatory variables. These results affirm the independence of predictors—EG, URB, EDU, DI, and EPS—ensuring that the estimated coefficients are not distorted by linear dependencies. This diagnostic supports the validity and reliability of the subsequent regression analyses, particularly the robustness of the Method of Moments Quantile Regression (MMQR) results.

Building on the stable multicollinearity profile, the MMQR estimates reported in Table 9 provide nuanced insights into how the effects of the independent variables on SE differ across its conditional distribution, capturing the heterogeneity among BRICS nations. In the linear model, URB consistently shows a strong positive and significant effect across all quantiles, indicating that urbanization—when accompanied by adequate infrastructure and governance—can support sustainability outcomes. This finding aligns with Bibri et al. (2020), who found that compact urban design and public service access can enhance ecological performance in emerging economies.

In contrast, EDU displays a significant negative effect across all quantiles, a finding that may initially seem counterintuitive. This result, however, echoes the paradox reported by Chen et al. (2024), who observed that educational expansion in the absence of environmental content may not yield eco-conscious behavior. Torroba Diaz et al. (2023) found that environmental literacy positively affects students’ sustainability actions, but general education alone does not foster eco-conscious behavior unless specific environmental education programs are implemented. In many BRICS countries, education systems have prioritized economic competitiveness and human capital development, often at the expense of ecological literacy. As a result, increased educational attainment or expanded enrollment—particularly at the secondary and tertiary levels—has not always translated into environmentally conscious behavior or policy engagement. While this interpretation remains partly speculative, it is worth noting the case of India, where expanding access to formal education has occurred alongside rising emissions. This trend is largely attributable to the nation’s reliance on coal and other fossil fuels for energy production. Nonetheless, studies have highlighted curriculum gaps in sustainability education that may limit the environmental impact of educational progress (UNESCO, 2025). Such patterns illustrate the complex and potentially nonlinear relationship between education and environmental outcomes proposed in H1.

DI also shows a consistently negative and significant effect on SE, particularly in lower quantiles, suggesting that digital infrastructure expansion may initially strain environmental resources. This is consistent with findings by Alsanie (2025), who reported that digitalization often increases electricity consumption and e-waste before efficiency benefits materialize. These results lend support to H2, highlighting the complex and potentially adverse impact of digital innovation on sustainability in its early stages of adoption. In contrast, E.G., and EPS demonstrate mostly negative or insignificant effects, indicating that economic expansion and environmental policy stringency may have context-specific or delayed impacts. Similar results are found in Efayena and Olele (2024), which emphasizes that policy effectiveness depends on enforcement capacity and institutional quality. These findings offer empirical grounding for H4, which proposes that the effects of economic growth and policy stringency on sustainability are limited and highly contingent on broader policy integration.

The nonlinear model deepens these insights by incorporating EDU², which reveals a significant U-shaped relationship between education and sustainability. While EDU remains negative, EDU² turns positive and significant at the median and upper quantiles. This confirms H1 by demonstrating a threshold effect, where education begins to foster sustainability only after surpassing a certain level of depth and quality. This finding supports the “threshold hypothesis,” where education only begins to promote sustainability after reaching a critical level of environmental integration and awareness. This is reinforced by Maneejuk and Yamaka (2021), who found similar nonlinearity in the role of education in East Asian economies. The case of China is illustrative here—where the national curriculum has increasingly integrated green competencies in recent years, resulting in measurable environmental improvements (Yang et al., 2022).

Moreover, DI in the nonlinear model shows diminishing negative effects, becoming insignificant at higher quantiles, suggesting that digital innovation may transition from a source of environmental burden to neutrality or benefit as supporting infrastructure and regulatory frameworks mature. This further supports H2, illustrating that the environmental effects of digital innovation evolve with a country’s technological maturity and policy environment. URB retains its strong positive influence across all quantiles in both models, reinforcing H3 and highlighting the importance of guided urban development as a stable contributor to ecological improvement. This echoes the findings of Goel et al. (2024), who argue that the environmental impact of digital tools evolves with a country’s digital maturity and policy safeguards. URB retains its strong positive influence across all quantiles in both models, highlighting the crucial role of managed urban development.

Overall, these results justify the use of the MMQR framework by revealing the non-uniform nature of relationships across different sustainability levels (see Figure 2). The nonlinear model, by capturing threshold effects of education and evolving digital impacts, provides a more refined and realistic understanding of the dynamics at play—offering valuable insights for policy design tailored to the specific sustainability profile of each BRICS nation.

The results from the Feasible Generalized Least Squares (FGLS) estimation, presented in Table 10, serve as a robustness check to validate the findings from the MMQR models. Overall, the FGLS outcomes largely corroborate the MMQR results, confirming the direction and significance of key relationships, while providing additional support for the stability of the core findings.

In both the linear and nonlinear specifications, URB maintains a positive and highly significant association with SE, reinforcing the conclusion that urbanization, when well-managed, contributes positively to environmental sustainability across BRICS countries. This consistency across models highlights urban planning and infrastructure development as a pivotal policy lever.

EDU remains strongly negative and significant in both models, consistent with the MMQR findings at lower quantiles. This reinforces the interpretation that mere enrollment—without emphasis on environmental curricula or eco-literacy—may not foster sustainability and could even align with increased consumption or industrial activity. However, in the nonlinear model, while EDU² is positive, it is not statistically significant, suggesting that the curvilinear (U-shaped) pattern observed in MMQR is more visible at distributional extremes and not uniform across the panel.

DI also continues to show a significant negative effect, aligning with MMQR findings at lower quantiles. This underlines that digital innovation, in its current form within BRICS, may exacerbate environmental pressures unless paired with green technology initiatives. The influence of, E.G., and EPS remains insignificant, confirming their limited direct impact on SE across the panel, which echoes the MMQR’s weaker and less consistent results for these variables.

The significant Wald test values in both models confirm the joint significance of the explanatory variables and the reliability of the estimates.

The Granger causality test results in Table 11 provide important insights into the directional relationships between SE and its predictors, revealing asymmetries in how different variables influence sustainability outcomes across BRICS nations. The bidirectional causality observed between DI and SE suggests a complex interplay between technological advancements and environmental sustainability. The statistically significant result for SE → DI (F = 7.221, p = 0.003 ) indicates that environmental conditions strongly dictate the trajectory of digital adoption, possibly through policy interventions aimed at mitigating the ecological impacts of digital expansion. Conversely, the weaker significance of DI → SE (F = 2.634, p = 0.090 ) implies that digital innovation exerts a marginal influence on environmental sustainability (Fang et al., 2023; McAleer, 2021; Tiwari et al., 2021; Yıldız et al., 2023), aligning with previous findings that excessive reliance on digitalization without robust green technology frameworks contributes to ecological strain rather than sustainability improvements.

EDU exhibits a unidirectional causality towards SE (F = 0.759, p = 0.047 ), suggesting that educational expansion drives environmental outcomes but not vice versa. This supports the earlier MMQR and FGLS results, where education was found to degrade the environment at lower quantiles, yet contribute positively when modeled nonlinearly. The absence of causality in the opposite direction (SE → EDU, F = 0.420, p = 0.661 ) further reinforces that environmental sustainability does not significantly alter educational structures, implying that any improvements in environmental consciousness through education must be proactively implemented rather than naturally evolving in response to environmental changes.

EPS fails to Granger-cause SE (F = 0.235, p = 0.792 ), nor is it influenced by SE (F = 1.424, p = 0.258 ), suggesting that regulatory frameworks alone do not substantially dictate sustainability outcomes. This finding resonates with MMQR results, where EPS displayed inconsistent significance across quantiles. The lack of causal directionality implies that without robust enforcement and systemic policy integration, environmental regulation on its own does not lead to measurable improvements in sustainability, reinforcing the need for complementary mechanisms such as technological innovations, educational reforms, and urban infrastructural development.

E.G., demonstrates a weak but significant causality towards SE (F = 0.883, p = 0.083 ), supporting the observation that economic expansion plays a role in influencing environmental outcomes, although not strongly. The lack of causality in the reverse direction (SE → E.G., F = 0.287, p = 0.753 ) suggests that sustainability conditions do not substantially alter economic growth trajectories, which is consistent with MMQR results indicating mixed effects of, E.G., on SE across quantiles. This highlights that while economic progress can sometimes align with environmental improvements, it is not an inherent driver and proactive policy measures are necessary to balance growth with ecological concerns.

URB presents bidirectional causality with SE, reinforcing the earlier MMQR and FGLS findings that structured urban expansion contributes positively to sustainability. The causality of URB → SE (F = 1.010, p = 0.065 ) indicates that increasing urban development fosters environmental improvements, likely through better infrastructure, green urban planning, and technological innovation. Meanwhile, SE → URB (F = 1.547, p = 0.038 ) suggests that sustainability conditions also influence urbanization strategies, possibly encouraging environmentally conscious urban policies in regions experiencing ecological strain. This bidirectional relationship underscores the interdependent nature of urbanization and sustainability, highlighting the importance of sustainable urban policies in shaping long-term environmental outcomes.

Overall, these findings reinforce key conclusions from MMQR and FGLS while providing deeper insight into the directional influences among variables. The unidirectional causality from education to sustainability validates the threshold-dependent impact of education found in nonlinear models, while the bidirectional causality between urbanization and sustainability affirms its role as a stabilizing factor in environmental outcomes. The weak causality from economic growth and the insignificance of environmental policy stringency signal that growth-oriented policies alone do not guarantee sustainability improvements, necessitating a broader focus on regulatory enforcement, infrastructure, and technological integration. Additionally, the asymmetric relationship between digital innovation and sustainability highlights the need for green digital transformation strategies, ensuring that technological advancements complement rather than counteract sustainability efforts. These causality findings offer valuable policy implications, emphasizing the need for targeted interventions that prioritize education, urban infrastructure, and technological sustainability while balancing economic growth with environmental considerations.

This study investigates the determinants of environmental sustainability (SE) across BRICS nations from 1990 to 2023, focusing on economic growth (EG), urbanization (URB), education (EDU), digital innovation (DI), and environmental policy stringency (EPS). Using a multi-method econometric approach—including Method of Moments Quantile Regression (MMQR), Feasible Generalized Least Squares (FGLS), and Granger causality analysis—we examined how these factors influence SE across varying levels of sustainability performance.

The MMQR results reveal that the influence of these variables is not uniform across the distribution of SE outcomes, indicating the presence of substantial heterogeneity among BRICS nations. Urbanization consistently exhibits a positive and statistically significant effect across most quantiles, suggesting that when managed effectively, it can contribute constructively to environmental outcomes. Education shows a more complex relationship: while associated with negative effects at lower quantiles, a U-shaped pattern emerges, indicating potential benefits at higher educational thresholds—highlighting the importance of both educational quality and curriculum content. Digital innovation also exhibits a generally negative relationship with SE at lower quantiles, suggesting that its benefits are conditional on policy context and technological maturity. Economic growth and environmental policy stringency yield mixed and weaker results, implying that these variables alone may be insufficient drivers of sustainability in this context.

Granger causality analysis further underscores the directional dynamics among variables. Education is found to Granger-cause SE, reinforcing its role as a key input to sustainability transitions. Urbanization and SE show bidirectional causality, suggesting feedback loops between urban development and environmental conditions. Digital innovation and SE exhibit asymmetric causality—where SE influences DI more than the reverse—highlighting that environmental priorities may drive technological change, rather than technology inherently producing sustainable outcomes. The weak or insignificant causality between SE and both economic growth and policy stringency supports the interpretation that deeper structural reforms are needed beyond economic expansion or regulation alone.

Taken together, these findings contribute to a more nuanced understanding of sustainability determinants in emerging economies. The patterns identified offer valuable empirical insights into how education, urbanization, digital innovation, and institutional variables interact within the BRICS context. As these insights are shaped by differences in governance structures, development trajectories, and policy environments, their relevance is most applicable to countries with similar socio-economic and institutional conditions. Broader generalizations may require further investigation in additional regional or global contexts.