Structural and Investment Determinants of Green Growth in Transition Economies: Evidence from South East and Eastern Europe

From the time of industrial evolution, the world has experienced robust economic growth, thus leading policymakers and researchers to study how economic development affects environmental sustainability [46].

In addition, several studies show that human activities directly contribute to environmental degradation [23]. Organizations such as the Intergovernmental Panel on Climate Change (IPCC) [9] and the United Nation states that economic activities are major causes of environmental degradation [47]. International bodies, institutions, specialists and practitioners increasingly emphasize the key role of green growth within the broader agenda of sustainable development. Green growth extends far beyond environmental concerns as it focuses on reshaping how economies expand. As nations rethink their development models, the durability and long-term resilience of their economies have become central issue [48].

Sustainable green growth requires clear conceptual foundations and consistent measurement systems. In response to this fragmentation, international bodies collaborated to establish shared knowledge platforms and indices to track progress. Since then, many countries particularly in the emerging economies have integrated green growth into national development strategies, with receiving recognition and supporting economic planning with environmental sustainability and social inclusion [49]. In addition, Sadorsky [50] found that the excessive use of natural resources as a result of economic expansion can harm the environment, increase production, the energy consumption and transport. Therefore, global organizations and researchers are placing greater emphasis on promoting environmentally sustainable economic activities [51]. However, it is worth noted that many countries continue to face difficulties in balancing economic growth with environmental protection and the sustainable use of natural resources [49, 52]. As a result, the idea of green growth has developed to promote economic growth while improving environmental quality [53].

Green growth aims to support steady economic progress while lowering the use of natural resources and decreasing carbon emissions. In addition, there is also the need for resilient economic, industrial, and financial systems. In this regard the environmentally friendly technologies play a key role in the process, encouraging green businesses and innovation. However, investing in green technologies involve high uncertainty, thus specialized financial tools and risk management strategies are needed to support such investments. As new technologies change how goods are produced and how markets operate, they can drive broader economic transformation. For this reason, the green growth approach considers both economic and financial factors [54].

Moreover, green growth offers integrated solutions to climate and development challenges, supports economic diversification, accelerates technological progress and promotes sustainable growth across sectors [55].

Since the Rio Summit, the global economy has struggled to balance continued economic expansion with growing environmental degradation. Green growth emerged as a framework to reconcile this competing pressure by promoting economic development while preserving natural resources essential for human well-being [10]. In addition, economic growth can reduce emissions through improved efficiency and cleaner technologies [56, 57] while others show that growth initially increases pollution until a certain threshold is reached [58]. Additional research highlights the role of education, trade openness, investment and regional differences in shaping environmental performance alongside growth [59]. According to Talebzadehhosseini and Garibay [60], green products promote green growth across 61 countries, while technological innovation itself shows no significant effect. In contrast, Mensah et al. [61] reported that sector-specific technological progress support green growth, particularly in transport in Ocean countries, manufacturing in Asia, and energy technologies in Europe. Further studies indicate the human capital closely linked to green technologies and regulation. In this regard the education is shown crucial factor towards enhancing the green growth and the green transition [28] and human capital plays a significant role in G7 countries [62]. While Lee and Kim [63] highlighted that this effect depends on the size and environmental awareness of the urban population.

Beyond these areas, fewer studies examine green growth’s link with industrial and labor outcomes [64], environmental regulation [65, 66], and political economy [67].

In addition, recent studies on finance-sustainability have traditionally focused on energy use and carbon emissions, but recent works increasingly emphasize green growth, which integrated economic expansion with environmental sustainability [68]. Furthermore, green growth-oriented studies draw on growth theory to assess whether sustainable growth is feasible in both theory and practice [15]. In addition, income levels, research and development spending, fiscal and monetary policies, and foreign direct investment are shown to play an important role in promoting green growth across different country groups [69, 70]. Macroeconomic stability also supports green growth, as evidenced in the EU, India and BRICS-T economies [47]. Evidence on financial development is mixed, because while it can foster green growth through innovation spillovers and regional effects, it may also delay local green growth when financial expansion is unbalanced [71, 72].

Moreover, to expand their economies, countries increasingly rely on FDI and trade, which can enhance technology transfer, human capital and productive efficiency. However, their environmental effects remain contested. The pollution haven hypothesis argues that trade and FDI shift pollution-intensive activities to host countries, worsening environmental quality [73], a view supported by evidence linking FDI to higher emissions [74, 75] though long-run effects may be insignificant [76]. Trade effects also differ, with exports increasing emissions and imports reducing them in some cases [77].

In contrast, the pollution hypothesis suggests that internationalization improves environmental performance through technology transfer and better practices, showing that trade and FDI can narrow technology gaps and enhance environmental quality under appropriate conditions [78]. Nonetheless, outcomes depend on policy and institutional quality, as effective regulation and innovation are needed to direct FDI toward greener activities [79].

Furthermore, green investment, credit and institutional quality improve environmental and economic outcomes across regions [59, 80]. Overall, the literature highlights the central role of financial development, innovation and human resources in green transition, though existing studies remain concentrated on single countries such as China, leaving cross - country evidence in emerging economies relatively limited [81].

This section presents the results obtained from econometric models and diagnostic tests. Table A in the Appendix provides the pooled OLS regression results that assess the impact of the explanatory variables on NETs. The results indicate positive and statistically significant effect of GCF (β = 0.823, p < 0.001). This implies that higher levels of investments positively may impact the sustainable economic growth or wealth creation after accounting for environmental costs measured with NETs. Additionally, the GDP per capita is found to have a positive and significant impact on NETs (β = 0.00129, p = 0.001). Moreover, industrial value added is also positively and significantly related to NETs (β = 0.403, p < 0.001).

In contrast, Foreign Direct Investment (FDI) has a negatively and significantly strong relationship (β = -0.735, p < 0.001), which indicates that FDI flows can possibly be unrelated to the positive contribution towards sustainable growth in the selected countries. On the other hand, unemployment has a significant and negative impact on NETs (β = -0.282, p = 0.004). Renewable energy consumption has a positive impact with marginal significance (β = 0.136, p = 0.054). This implies that renewable energy consumption may have a positive impact on sustainable development. Otherwise, exposure to PM2.5 emissions (β = 0.178, p = 0.169) and trade openness (β = 0.048, p = 0.164) do not show statistical significance under the study circumstances. Table A presents the Pooled OLS regression results.

Table B in the Appendix provides the results of the RE estimation of the determinants of adjusted net savings (NETs). It should be noted that the estimated coefficients are still very stable with respect to the values that were estimated using the pooled OLS model, which indicates that the direction and significance of relations are rather consistent and robust under different estimators used. Specifically, such explanatory variables as gross capital formation (β = 0.823, p < 0.001), GDP per capita (β = 0.00129, p = 0.001), and industrial value added (β = 0.403, p < 0.001) still have a positive and statistically significant impact on NETs, while foreign direct investments (β = -0.735, p < 0.001) and unemployment (β = -0.282, p = 0.003) have a negative and statistically significant impact.

Renewable energy consumption is still positively related to NETs (β = 0.136, p = 0.052), although the statistical significance is marginal under the RE model. On the other hand, PM2.5 exposure (β = 0.178, p = 0.167) and trade openness (β = 0.048, p = 0.162) are not significant. High consistency between the two models is explained by the minimal between-group variance estimate (rho = 0) in the RE model. To address issues of potential within-country correlation and heteroskedasticity, the RE model has been re-estimate with cluster-robust standard errors clustered at the country-level (8 clusters) presented under Table C.

Table C presents RE estimates using cluster-robust standard errors. It is seen that gross capital formation (β = 0.823, p < 0.001) and industrial value added (β = 0.403, p = 0.001) are still positive and statistically significant factors for the NETs, whereas the relationship between foreign direct investment and NETs is still negative and highly significant (β = -0.735, p < 0.001). GDP per capita is found to have a positive impact on NETs, marginally significant (β = 0.00129, p = 0.070). On the other hand, PM2.5 (β = 0.178, p = 0.597), trade openness (β = 0.048, p = 0.432), unemployment rate (β = -0.282, p = 0.149), and renewable energy (β = 0.136, p = 0.429) have no statistically significant effect on NETs in the context of clustered robustness test correction. The interpretation of the insignificant relationship between PM2.5 exposure and NETs needs further examination due to possible lags in environment impacts, measurement issues, and adjustments for sustainable NETs.

In order to find out which panel specification to use, a Hausman test was carried out to differentiate between the RE and FE models. The null hypothesis is that the difference between the parameters is random, i.e., the RE model provides consistent and efficient results. The alternative hypothesis indicates that RE results are inconsistent because of correlation between individual effects and regressing variables. The results reject the null hypothesis because χ²(8) is greater than 38.58 and p-value < 0.001. This shows that the differences between FE and RE models are statistically significant. This suggests that the model assumption that unobserved country effects are uncorrelated with regressing variables is violated.

The results make sense in view of the large correlation established in the previous results between the individual effects and the regressors, i.e., (corr(u_i, Xb) = -0.7792). Therefore, the results from the FE estimator are consistent and preferred. The size of the differences in the coefficients also adds credence to the above finding. Notable differences arise for the coefficients of GDP pc, industry share, FDI, and unemployment levels, warranting the notion that country-specific factors are systematically related to these variables. Also, the reversal in the signs of the coefficients of the unemployment levels and trade openness between the FE and RE regression results points towards the presence of significant differences across the countries. Although the Hausman test results summary indicates that the variance difference matrix (V_b − V_B) is not positive definite—which is common in a finite sample case—the chi-square results are very significant in favor of the fixed-effects model.

From the Hausman test results, the FE is suggested over the RE, this is because the RE estimator appears inconsistent due to correlation of unobserved country-specific factors with the regressors. In addition, Table D presents FE results, and Table E shows Hausman test results

4.1 Model selection and reliability assessment

The result of rejecting the null hypothesis verifies our line of argument that conventional FE or RE standard errors are not fully reliable under the study circumstances. Accordingly, using cluster robust standard errors is more appropriate for consistent results. To address issues of heteroskedasticity, serial correlation, and cross-sectional dependence, the FE model is further estimated using standard errors following the Driscoll-Kraay approach with a slight modification of the maximal lag equal to two (Table F). This is especially appropriate in a macro panel data environment featuring a lower number of cross-section units, i.e., N = 8. The results continue to highlight that the Fixed Effect Driscoll-Kraay is statistically significant overall (F(8,20) = 320.49; p < 0.001), with the R² remaining unchanged (=0.6317).

It is worth emphasize that based on the econometric results obtained from the Fixed Effect Driscoll-Kraay, chosen between pooled OLS and RE, as appropriate model under the study circumstances, the statistically significant variables are GDP per capita, gross capital formation, industrial value added and renewable energy consumption and have a positive effect on green growth as measured by Adjusted net savings, including particulate emission damage (% of GNI). In contrast, foreign direct investment has a negative effect on green growth. Otherwise, the coefficients of PM2.5 emissions, trade openness and unemployment remain statistically insignificant under the conditions of this study presented in Appendix Table F. FE regression with Driscoll-Kraay standard errors.

To determine the existence of multicollinearity among the independent variables, the results of VIF test were obtained as presented in Table 4. The results have shown that the values of VIF for the independent variables are very low, far below the critical value of 10, and even below the rather conservative critical value of 5 as proposed in the field of panel econometrics. The maximum value of VIF is found in the variable of renewable energy consumption (VIF = 4.87) and exposure to PM2.5 (VIF = 4.13). Although the values of VIF hint towards moderate correlations with other regressors for these two variables, the values are nevertheless within acceptable limits and do not hint at severe multicollinearity. The other variables have even lower VIF values: Unemployment (3.49); Industry share (2.98); Gross capital formation (2.38); Trade openness (2.18); FDI (2.15); GDP per capita (1.52). The average VIF value is 2.96. This is well within tolerance levels. Hence, we can conclude that the levels of correlation between the explanatory variables are not high. Also, we conclude that multicollinearity will not impact the estimates of the regression coefficient. Overall, the diagnostic evidence clearly verifies that multicollinearity is not an essential problem that the model faces, hence boosting the credibility of the results obtained by the model.

Table 4. Variance inflation factors for the explanatory variables

Table 5. Wooldridge test for autocorrelation in panel data

Table 5 presents the results obtained from Wooldridge test for autocorrelation in panel data. The results suggest to reject the null hypothesis (F(1,7) = 22.440, p = 0.0021), and indicate the presence of serial correlation in the panel data under the study circumstances thus justifying the use of robust estimation techniques.

Table 6 presents the OLS Robust, FE, and RE estimations for the factors determining NETs. The coefficients in the three estimators for the variable’s gross capital formation and industrial value added have remained positive, whereas that for the GDP per capita has been found to be positive and significant in all three estimations. On the other hand, the coefficient for FDI reveals negative signs on sustainability indicators, which indicates a mixed effect of FDI in transition economies.

The coefficients of the two variables PM2.5 and trade openness are not found to be statistically significant in the three estimations. Nevertheless, it cannot be inferred from this that pollution does not affect sustainable development, considering that the variable itself represents the sustainability-adjusted index, and environmental issues might take time to influence such an economic indicator.

The same reasoning may also apply to renewable energy consumption, as it is not found to be statistically significant in some cases.

Table 6. Pooled ordinary least squares (OLS) with robust standard errors, fixed effect (FE), and random effects (RE) results

Table 7. Summary of hypothesis testing results

Table 7 presents the summary of the hypothesis testing and results. Most of the hypotheses raised in this study find strong support by the empirical findings. Gross capital formation is the most influential determinant of green growth performance, confirming that investment intensity plays a binding role in sustaining adjusted net savings.

Similarly, GDP per capita exerts a positive and statistically significant within-country effect, which supports income growth as enhancing sustainable capital accumulation over time. Industrial value added follows with a strong and quite consistent positive contribution, suggesting that structural transformation is an important factor shaping green growth dynamics in transition economies.

The results from the estimation of the FDIs relationship with adjusted net savings reveal a statistically significant but with negative impact, which supports the hypothesis of the association of FDIs with ecologically intensive types of economic activity within the economies under investigation.

Renewable energy consumption presents a positive effect, which reveals statistical significance when using robust estimation of the Driscoll-Kraay estimation. The results offer partial verification of the support of such an economic factor in the promotion of green growth. The effect of exposure to PM2.5 fails to reveal statistical significance; thus, the associated hypothesis is rejected.

The results support the dominant explanation of green growth in transition economies, highlighting the investment processes and economic factors alongside with renewable energy, as key drivers within this economic development model.

The determinants of green growth in the selected transition economies of South East and Eastern Europe including Kosovo, Albania, Bosnia and Herzegovina, North Macedonia, Serbia, Montenegro, Moldova, Ukraine, Georgia, Armenia, and Azerbaijan with the data from WDI database for period (2000-2024) have been examined in the current study by the use of the panel data methods. The econometric results indicate that green growth, represented by the adjusted net savings, is driven by the investment intensity, income growth, and industrial composition, whereas the controls of trade, pollution, as well as labor, do not offer robust results. The FE model, which is verified by the Hausman test, indicates that gross capital formation has the strongest and most consistent impact on the determinants of green growth. Higher rates of investment are observed to strengthen the rates of adjusted net savings in countries over time, thereby emphasizing the importance of capital accumulation and the need to sustain economic wealth in the long run. The within-country estimates are appreciably positive and statistically significant when measuring the effects of GDP per capita on sustainable savings performance. In addition, the positive and consistent effects of industrial value added can be attributed to the potential positive impacts of structural change towards the industrial sector to capital deepening, rather than negatively affecting sustainability. In contrast, the results for foreign direct investment reveal a negative sign, which is significant for both the FE and robust estimations. It can therefore be implied that the inflows of FDI, as far as the transition countries included in the study are concerned, may be connected with environmentally intensive activities, contributing to decreased net savings. PM2.5 exposure and trade openness do not present any directly significant effects, while the impact of renewable energy consumption is positive, although it is significant after the application of the correction suggested by Driscoll and Kraay. Overall, what these findings suggest is that green growth in transition economies is largely driven by investment composition and structural economics, rather than openness or external investment. The sustainable development strategy in transition economies should place greater emphasis on enhancing productive investment and the environmental quality of investment, as well as promoting structural change in transition economies, especially in industrial efficiency.

The findings of this study indicate that green growth in transition economies depends on the quality and composition of economic transformation, highlighting that domestic capital formation, industrial development and renewable energy developments contribute positively to regulated net savings, while foreign direct investment exhibits a negative relationship. Among other things, it is worth noting that not all forms of investment equally support sustainable wealth creation. Therefore, it is important to guide economic transformation towards productive, environmentally responsible and long-term activities. Green growth should be seen not only as a consequence of economic growth, but as the result of a development model that aligns investment decisions, industrial modernization and energy transition with sustainability objectives.