Financial risk and renewable energy: exploring the influence of urbanization and natural resource rents across 112 countries

Unit root test

Before performing the regression, we need to test the smoothness of each variable. For the characteristics of our selected data, we chose four methods to test the smoothness of the data: the IPS (Im Pesaran Shin) test, the HT (Hadri) test, the Fisher-ADF (Fisher-Augmented Dickey-Fuller) test, and the Fisher-PP (Fisher-Phillips-Perron) test. The test results are shown in Table 5. According to the test results, it can be seen that all variables are smooth at the first-order difference. The variables passed the data smoothness test and met the prerequisites for the following tests and regressions.

Cointegration test

After determining that the data are smooth, a cointegration test is required. The purpose of the cointegration test is to verify whether there is a long-run cointegration relationship between the variables. We chose three methods to conduct the cointegration test: the Kao test, the Pedroni test, and the Westerlund test. The test results are presented in Table 6. According to the test results, all nine test statistics of the three methods reject the original hypothesis of “no co-integration” at the 1% level. This indicates that there is a long-term stable equilibrium relationship between renewable energy consumption, financial risk, GDP, trade openness, urbanization, and natural resource rent. This proves that the next step of the empirical regression analysis can be carried out.

Benchmark regression analysis

To verify whether financial risk has a significant impact on renewable energy, we conducted a benchmark regression empirical analysis. The regression results are shown in Table 7. Column (1) of the table shows the regression results without adding control variables but controlling for individual and time effects. Column (2) demonstrates the regression results with the inclusion of control variables but without controlling for individual versus time effects. Column (3) shows the regression results with the inclusion of control variables and controlling for individual and time effects. It can be found that financial risk has a significant dampening effect on renewable energy consumption regardless of whether control variables are included or not and whether individual and time effects are controlled. We know that individual and time differences will have an impact on the regression. Therefore, we should include two-way fixed effects in the regression to eliminate this potential effect. Therefore, the regression results in column (3) can be used for detailed analysis. The regression coefficient for financial risk is –0.414 and is significant at the 5% level. This indicates that a 1% increase in the level of financial risk reduces the level of renewable energy consumption by 0.414%.

This finding is consistent with Ahmad et al. (2022), who explored the impact of financial risk on renewable energy initiatives. Their research highlights that financial instability can constrain budgets for renewable energy technologies, which are vital for fostering technological innovation and enhancing long-term renewable energy consumption. Similarly, Sweerts et al. (2019) emphasized the importance of financial de-risking as a critical measure to unlock the potential of renewable energy, particularly in Africa. In addition, Liu et al. (2023) examined the role of financial development in facilitating the transition to renewable energy and found that, over the long term, financial development promotes sustainable energy transitions, further supporting our conclusions.

The underlying reason is that as financial risk rises, investors demand higher returns to compensate for the increased risk, leading to higher financing costs (Khan et al., 2022). Renewable energy projects often require significant upfront investment and have long payback periods, making higher financing costs unaffordable for some investors. This limits capital inflows and hampers the progress of projects, ultimately constraining the further development of renewable energy (Wang et al., 2024c). Additionally, as financial market volatility increases, investors’ expectations for the long-term returns of projects become more uncertain. Given that renewable energy projects typically take years or even decades to break even, increased financial risk may cause investors to doubt the future returns of these projects. This uncertainty pushes them toward short-term or lower-risk investment options, thereby reducing investment in renewable energy projects.

As can be seen from the table, the regression coefficient for GDP is significantly negative (-2.04), while the regression coefficient for GDP² is significantly positive (0.113). This shows that the effect of GDP on renewable energy consumption is a “U” shape. When economic development is at a low level, economic growth increases the proportion of traditional energy use, slowing down the pace of energy transition. However, when the economic level grows to a certain level, the promotion effect of economic growth on the use of renewable energy begins to appear, at which time economic growth contributes to the energy transition.

The regression coefficient for trade openness is 0.229 and is significant at the 10% level. This indicates that a 1% increase in trade openness increases the level of renewable energy consumption by 0.229%. Increased openness facilitates the introduction and exchange of renewable energy technologies and accelerates the energy transition (Kim and Kim, 2015).

The regression coefficient for urbanization is –1.338 and is significant at the 1% level. This indicates that a 1% increase in the level of urbanization reduces the level of renewable energy consumption by 1.338%. In addition, the effect of natural resource rent on energy transition is not significant.

Robustness test

To further validate the robustness of the impact of financial risk on renewable energy consumption, this study employs three methods for verification, as presented in Table 8.

(1) There may exist a bidirectional causal relationship between financial risk and renewable energy consumption, along with the potential for omitted variable bias or simultaneity issues within the model, which could lead to biased estimates in traditional regression methods. To address these endogeneity concerns, the study first employs the system GMM model to re-examine the relationship between financial risk and renewable energy consumption. The regression results obtained using this approach are presented in Column (1).

(2) Given the possibility of a long-term stable relationship between financial risk and renewable energy consumption, the FMOLS model is utilized. This method provides robust estimates in the presence of cointegration relationships, thereby avoiding potential biases inherent in traditional OLS regressions. To enhance the reliability of the robustness check, the FMOLS model is employed as an alternative, with the results presented in Column (2).

(3) To evaluate the robustness of the conclusions over varying time spans, the sample period is adjusted to 1999–2019, narrowing the time frame to exclude certain external shocks that may have influenced the results. By conducting regression analysis over this shorter time window, the study assesses whether the relationship between financial risk and renewable energy consumption remains consistent, ensuring the robustness and generalizability of the findings. The new estimation results are reported in Column (3).

The results from all three robustness checks are largely consistent with the baseline regression findings, confirming that financial risk significantly reduces renewable energy consumption. This consistency underscores the robustness of the conclusions drawn in this study.

Moderating effects analysis

To further explore the heterogeneous effects of financial risk on renewable energy consumption under different socio-economic contexts, we constructed a moderated effect model incorporating urbanization and natural resource dependence as moderating variables (Eqs. (2)–(4)). Specifically, urbanization influences energy demand and consumption patterns, while also affecting the role of financial risk by altering local fiscal capacities and market structures. Additionally, natural resource dependence determines the stability of regional economies and the sensitivity of financial risk. The regression results are presented in Table 9.

The results of the moderated effects regressions are shown in Table 9. The first column shows the regression results when the moderating variable is urbanization (Urb). It can be observed that the direction of the regression coefficients for each variable remains the same as the baseline regression. In addition, the coefficient of the interaction term is –3.061 and significant at the 5% significance level. The coefficient of financial risk on renewable energy consumption can be calculated as (−0.47-3.061*lnUrb). This means that the effect of financial risk on renewable energy consumption changes with the change in urbanization level. In other words, an increase in the level of urbanization worsens the dampening effect of financial risk on renewable energy consumption. This finding suggests that as urbanization increases, the dampening effect of financial risk on renewable energy consumption becomes more pronounced. This observation aligns with Li et al. (2021), who documented how excessive financial support during China’s rapid urban expansion exacerbated fiscal pressure and financial risks. Similarly, Cai et al. (2019) emphasized that although China’s land-based fiscal model during urbanization initially delivered economic benefits, it also concealed significant financial risks, which intensified at the advanced stages of urbanization. These findings underscore the critical role of urbanization in shaping the financial feasibility of renewable energy projects.

The underlying reasons for this are twofold: First, urbanization is accompanied by population concentration and infrastructure expansion, leading to a rapid increase in energy demand. Highly urbanized areas require significant amounts of energy to sustain city operations, particularly in sectors like transportation and construction (Wang et al., 2024b). This makes cities more vulnerable to financial risk, as tight funding and rising financing costs reduce their capacity to invest in renewable energy. Furthermore, since renewable energy projects often require substantial upfront investment and have long payback periods, cities may struggle to secure sufficient funding for these projects when financial risks rise. Thus, the greater the level of urbanization, the higher the energy demand, and the more severe the negative impact of financial risk on renewable energy. Second, urbanization also introduces specific financing challenges. Financial resources in urban areas may be concentrated on short-term, high-return projects (such as real estate or tech companies), neglecting renewable energy projects that offer long-term environmental benefits. Risks in the real estate market and the over-concentration of capital can lead to a misallocation of financial resources, suppressing investment in renewable energy. Additionally, infrastructure projects and renewable energy developments often compete for funding, and in times of rising financial risk, investors and governments may prioritize projects with immediate returns, such as roads or public transportation, over renewable energy. Moreover, the policy uncertainties and social divisions that accompany urbanization further amplify the effects of financial risk, weakening the feasibility of renewable energy projects.

The second column presents the regression results when the moderating variable is natural resource rent (lnNS). As can be seen from the regression results in the second column, the direction of the coefficients of the variables is also consistent with the benchmark regression. In addition, the coefficient of the interaction term between financial risk and natural resource rent is 0.208 and significant at the 5% level. It can be concluded that the coefficient of financial risk on renewable energy consumption is (–0.425 + 0.208* lnNS). This again implies that the effect of financial risk on renewable energy consumption is moderated by natural resource rent. As natural resource rent increases, the dampening effect of financial risk on renewable energy consumption diminishes. This finding is consistent with Wang et al. (2025a), who argued that effective natural resource management could offset financial risks. Additionally, Alsagr (2024) emphasized the critical role of geopolitical risk and corruption in resource-rich countries, further contributing to the discussion in this field.

Resource-rich countries primarily buffer the impact of financial risk through the fiscal stability and flexibility provided by natural resource rents. First, stable natural resource rents offer strong fiscal support, enabling countries to maintain renewable energy investments even during periods of heightened financial risk (Wang et al., 2024a). These countries can leverage export revenues to sustain fiscal stability, thereby reducing reliance on external financing and mitigating the uncertainties caused by financial market fluctuations. Moreover, many resource-rich countries have established sovereign wealth funds and foreign exchange reserves. These funds not only help shield the country from financial risk but also support long-term sustainable development and renewable energy investments. Third, some resource-rich countries actively use resource revenues to drive their energy transitions. Recognizing the long-term risks of fossil fuel dependence, these nations allocate natural resource rents toward renewable energy investments, fostering both energy transition and sustainable economic development for the future.

The third column shows the empirical regression results when urbanization (lnUrb) and natural resource rent (lnNS) are both used as moderating variables. As mentioned above it can be observed that the regression coefficients of the variables are also consistent with the direction of the coefficients of the benchmark regression. In addition, the coefficient of the interaction term between financial risk and urbanization is significantly negative (–3.015) and the coefficient of the interaction term between financial risk and natural resource rent is significantly positive (0.203). This indicates that the effect of financial risk on renewable energy consumption is moderated by both urbanization and natural resources. At this point, the impact of financial risk on renewable energy consumption can be calculated as (-0.48-3.015*lnUrb + 0.203* lnNS).

Threshold effect analysis

Section “Moderating effects analysis” confirmed the moderating roles of urbanization and natural resource dependence in the relationship between financial risk and renewable energy consumption. Building on these findings, variations in urbanization and natural resource dependence may alter the extent to which financial risk impacts renewable energy consumption. To further analyze the trends in these moderating effects as urbanization and natural resource dependence change, we apply a panel threshold model (Eqs. (5)–(6)) to explore how different threshold intervals of these variables influence the relationship between financial risk and renewable energy consumption.

Existence test for threshold effects and determination of the number of thresholds

The first thing we need to test in this section is whether a threshold effect exists. If it exists, we need to determine the number of thresholds and use this to determine the model. The results are shown in Table 10. It can be found that when the threshold variable is urbanization, the double threshold test fails, but the single threshold test is passed. The F-statistic is 113.25 and is significant at the 10% level. Therefore, for urbanization, we are going to use a single threshold model.

Similarly, for natural resource rents, the double-threshold effect test is not passed, but only the single-threshold effect test is passed. The F-statistic of the single threshold test is 107.19 and it is significant at the 5% level. Therefore, for natural resource rents a single threshold model should be used for estimation.

Estimation of thresholds and confidence intervals

After identifying it as a single threshold model we need to estimate the threshold values. We follow the suggestion of Hansen (1999) and search only for non-repeated values of the threshold variables lnUrb and lnNS. These non-repeated values are sorted in ascending order, ignoring about 1% of the observations before and after, and using only the middle 98% of the samples as a candidate range for the threshold, i.e., searching in the interval (1%, 99%). To improve the accuracy of threshold estimation, we adopt the “grid search method” used by Hansen in threshold regression to give the threshold value \(\gamma\) in threshold regression. The range of threshold values is first gridded using 0.0025 as the gridded level. Then, all the grid points obtained after rasterization are used as candidate threshold values \(\gamma \,\). The model is estimated one by one, the residual sum of squares is obtained, and the threshold value corresponding to the smallest residual sum of squares of the model is selected, which is the threshold estimation value\(\,\hat{\gamma }\,\).

The estimation results are shown in Table 11. It can be found that the threshold value of urbanization is 4.32 with a confidence interval of [4.308, 4.328]. The threshold value of natural resource rent is 3.119 with a confidence interval of [3.037, 3.216]. To demonstrate the threshold construction process more easily, we plotted the likelihood function images (Fig. 2 and Fig. 3).

This figure presents the likelihood ratio function for urbanization (lnUrb), highlighting the variation in financial risk across different urbanization levels.

This figure shows the likelihood ratio function for natural resource rents (lnNS), demonstrating the relationship between financial risk and natural resource rents.

Threshold effect regression results

The regression results for threshold effects are presented in Table 12. To ensure the accuracy of the estimation, we include a time dummy variable to control for the time effect in the estimation process. In addition, the regression results of the time dummy variables are not shown in the table to ensure a concise presentation of the results. In the table, lnFR_0 indicates the regression coefficient of financial risk when the threshold variable is smaller than the threshold value, and lnFR_1 indicates the regression coefficient of financial risk when the threshold variable is larger than the threshold value.

The regression results are presented in the second column when urbanization is used as a threshold variable. We can see that when the level of urbanization is below the threshold value of 4.32, the coefficient of the impact of financial risk on renewable energy consumption is –0.431 and significant at the 1% level. This means that a 1% increase in financial risk at low levels of urbanization reduces the level of renewable energy consumption by 0.431%. Whereas, when the level of urbanization is above the threshold value of 4.32, the regression coefficient of financial risk on renewable energy consumption is –0.576 and significant at a 1% level, as shown in Fig. 4. This indicates that a 1% increase in financial risk at high urbanization levels results in a 0.576% decrease in the level of renewable energy consumption. The regression results in the first column indicate that differences in the level of urbanization do result in differences in the impact of financial risk. The impact of financial risk on renewable energy consumption is smaller at low levels of urbanization.

This figure reflects the threshold effect of financial risk on renewable energy development, with a focus on urbanization levels (lnUrb) as a threshold variable.

This finding aligns with previous studies. Yang et al. (2016) demonstrated that urbanization has stage-dependent effects, with the impact of financial risk on energy consumption varying across urbanization levels. They noted that highly urbanized areas, with their more complex economic structures and energy needs, are particularly sensitive to financial volatility. Similarly, Salim and Shafiei (2014) found that in OECD countries, urbanization exacerbates the vulnerability of energy consumption to financial instability. These studies suggest that urbanization amplifies the challenges posed by financial risk to renewable energy investment, particularly in regions with advanced urban development.

Yang et al. (2016) argued that highly urbanized regions, with their more complex economic structures and energy demands, are more susceptible to market disruptions, policy changes, and financial instability. This may explain why financial risk has a more pronounced dampening effect on renewable energy consumption in cities with higher levels of urbanization. In contrast, in areas with lower urbanization levels, energy demands are less complex, and the impact of financial risk on renewable energy consumption is less severe(Wang et al., 2025b). This is because lower urbanization typically results in simpler economic structures and lower dependence on high-cost, market-driven energy systems. Salim and Shafiei (2014) found similar patterns in OECD countries, where the impact of financial volatility on renewable energy consumption decreased in less urbanized areas.

The regression results are presented in the third column when natural resource rent is used as a threshold variable. Based on the regression results, we can find that when the level of natural resource rent is below the threshold value of 3.119, the regression coefficient of financial risk on renewable energy consumption is –0.508 and is significant at the 1% level. This means that a 1% increase in financial risk reduces renewable energy consumption by 0.508%. The regression coefficient of financial risk is –0.379 when the level of natural resource rents is above the threshold value of 3.119, as shown in Fig. 5. In other words, a 1% increase in financial risk suppresses renewable energy consumption by 0.379%. We can see that natural resource rent does have a significant threshold effect in the process of financial development on renewable energy consumption. And when the natural resource rent is at a high level, financial risk has a less deterrent effect on renewable energy consumption. This finding is consistent with Wang et al. (2024d), who argued that higher natural resource rents can buffer the negative impact of financial risk, particularly in energy transitions. Su et al. (2022) also pointed out that natural resource rents provide fiscal stability and enhance financial flexibility, helping countries manage the long-term investments required for renewable energy projects. The increased revenue from resource rents enables governments and enterprises to continue investing in renewable energy despite financial volatility.

This figure reflects the threshold effect of financial risk on renewable energy, with natural resource rents (lnNS) as the threshold variable.

Several factors may explain this relationship: First, natural resource rents provide governments with additional fiscal resources, which can be used to support renewable energy investment, subsidies, and favorable policies, even during periods of financial instability. (Su et al., 2022) emphasized that these rents can also attract international capital and technology, expanding financing opportunities for renewable energy projects. Furthermore, high resource rents can help countries diversify their financial risks, ensuring that renewable energy projects remain viable even in uncertain financial environments.

Overall, higher natural resource rents provide more policy flexibility, helping to alleviate the suppressive impact of financial volatility on renewable energy consumption.

As for the control variables, the regression coefficients of GDP in the second and third columns show very significant negative characteristics (–2.999, –2.914), while the regression coefficients of GDP² in both columns show significant positive characteristics (0.168, 0.161). Both of them show that the effect of GDP on renewable energy consumption has a “U” shape, which is consistent with the benchmark regression. The regression coefficients of trade openness are also positive at different significance levels (0.243, 0.196). Increased trade openness accelerates renewable energy consumption, which is also consistent with the benchmark regression.