Impact of risk factors on the link between natural resources rents and carbon emissions: Evidence from economic, financial, and political risks

The study addresses how revenues from natural resources (natural resources rents) influence carbon emissions across countries and how national risk conditions—economic, financial, and political—modify this relationship. Motivated by concerns over climate change, the carbon intensity associated with resource extraction, and heterogeneity across countries and emission levels, the authors use a quantile perspective to move beyond mean effects. They ask: (1) Do natural resources rents raise carbon emissions at different emission intensities? (2) How do economic, financial, and political risks moderate the rent–emissions link across the emissions distribution? (3) Are there threshold levels of these risks at which the effect of resource rents on emissions changes sign? Using 1990–2020 data for 66 countries, the study aims to inform sustainable natural resource management and risk management policies that can curb emissions related to resource exploitation.

The review highlights two strands. First, on natural resources rents and emissions, several studies find that dependence on natural resources increases carbon emissions and degrades environmental quality (e.g., Agboola et al. 2021 for Saudi Arabia; Huang 2022 for the US; Liu et al. 2022 using quantiles; Aladejare 2022 in rich African economies; Nwani et al. 2023). Some work indicates that effective management and clean energy transitions can improve ecological efficiency (Alfalih and Hadj 2022; Chen et al. 2022; Khan et al. 2022; Liu et al. 2023). Second, on risks and emissions: Economic risk/uncertainty can raise emissions by depressing innovation and altering energy use, though effects vary by time horizon and emission level (Adams et al. 2020; Yu et al. 2021; Cui et al. 2021; Syed et al. 2022; Anser et al. 2021; Chu et al. 2023; Adebayo et al. 2023a). Financial risk/conditions can reduce emissions via better financing for green projects, though financial development can also increase emissions depending on context (Pata et al. 2023; Hussain et al. 2022; Wang et al. 2023c; Ahmad et al. 2022, 2023; Qalati et al. 2023; Ling et al. 2022; Qin et al. 2021; Adebayo et al. 2023a). Political risk relates strongly to environmental outcomes: better institutions and stability generally enable mitigation, while high political risk and corruption undermine environmental quality (Su et al. 2021; Ahmad et al. 2023; Xia et al. 2022; Khan et al. 2021; Tang et al. 2022; Depren et al. 2023; Danish and Ulucak 2020; Dasgupta and De Cian 2018; Zhou et al. 2020; Chen et al. 2018; Li et al. 2023). Environmental regulations can mitigate the “carbon curse” (Fan et al. 2022; Che and Wang 2022; Yang et al. 2023; Su and Xu 2022). The review notes a gap: few studies jointly examine resource rents with economic, financial, and political risks, particularly the latter two.

Data: Panel of 66 countries from 1990–2020. Dependent variable: per capita CO2 emissions (metric tons per capita, WDI). Key independent variable: total natural resources rents (% of GDP: oil, gas, coal, minerals, forests, WDI). Moderators: economic risk (ICRG economic risk index), financial risk (ICRG financial risk index), political risk (ICRG political risk index). Higher ICRG scores indicate lower risk/greater stability. Controls: GDP per capita (2017 USD), urbanization (% urban population), industrialization (industry value added % of GDP), all from WDI. VIFs are below 5 (Ingdp 4.23; Inur 2.73; Inpr 2.36; Iner 2.32; Infr 1.98; Inrent 1.51; Inind 1.40), suggesting no multicollinearity. Variables are log-transformed to mitigate heteroscedasticity. Econometric approach: - Stationarity and cointegration: First-generation unit root tests (Fisher-PP, Fisher-ADF) and second-generation CIPS (Pesaran 2007). Most variables are stationary at first differences; InCO2, Ingdp, Inind non-stationary in levels but stationary in first differences; Pedroni panel cointegration tests reject no-cointegration at 1%, indicating long-run relationships. - Baseline effects: Panel quantile regression with nonadditive fixed effects (Powell 2022) across τ = 0.1–0.9 to estimate heterogeneous effects of natural resources rents on lnCO2, controlling for risks and macro covariates. - Moderation: Include interaction terms between lnrent and each risk index (lner, lnfr, lnpr) in quantile frameworks to assess moderation across the emissions distribution. - Robustness: Pooled OLS (mixed effects notation in paper), 2SLS for endogeneity, and system GMM (Arellano-Bover/Blundell-Bond) with AR(1)/AR(2) and Hansen tests. - Threshold effects: Dynamic panel threshold models (Seo and Shin 2016; Seo et al. 2019) with endogenous regressors and country fixed effects. Specification includes lagged lnCO2 and an indicator function I{lnrisk > γ} for each threshold variable (economic, financial, political risk). Estimation via GMM with grid search over γ; bootstrap tests assess linearity vs threshold nonlinearity; AR(1), AR(2), and Hansen J validate model assumptions. The study reports threshold values (with confidence intervals) and estimates of lnrent effects below/above thresholds.

- Stationarity/cointegration: Most variables are I(0) or I(1); lnCO2, lngdp, lnind become stationary in first differences. Pedroni tests indicate cointegration among variables (1% level). - Baseline quantile effects (Table 3): Natural resources rents increase lnCO2 across quantiles τ=0.1–0.9, with coefficients: 0.0629***, 0.103***, 0.0822***, 0.0629***, 0.0753***, 0.0704***, 0.0849***, 0.0624 (ns at 0.8), 0.170***. Effects are stronger at mid-to-high quantiles (≥0.07 from 0.5 to 0.9, excluding 0.8), indicating larger emissions responses in higher-emitting contexts. - Covariates in quantiles: • Economic risk (lner): Mostly negative effects on emissions (except τ=0.1 positive), strongest reductions at τ=0.8 and 0.9 (e.g., −0.152***, −0.463***). • Financial risk (lnfr): Generally negative association with emissions; insignificant at τ=0.4, 0.5, 0.8; significant reductions at other quantiles (e.g., −0.170*** at τ=0.3, −0.271*** at τ=0.9). • Political risk (lnpr, higher=lower risk): Positive and significant across all quantiles (≈0.172–0.502), implying greater political stability correlates with higher emissions in the main effect. • GDP per capita: Positive and significant across all quantiles (~0.44–0.73). • Urbanization: Positive at low quantiles (τ=0.1–0.2), negative and significant at higher quantiles (τ=0.3–0.9). • Industrialization: Mixed signs across quantiles. - Robustness (Table 4): • POLS Inrent coefficient 0.121**; 2SLS 0.1265***; system GMM 0.0136**; all indicate positive link between resource rents and emissions. • System GMM diagnostics: AR(2) p=0.180 (no second-order serial correlation), Hansen p=0.983 (valid instruments). - Moderation by risks in quantiles: • Economic risk interaction (Inrent×Iner, Table 5): Positive and significant at low quantiles (τ=0.1–0.4: 0.0492***, 0.0241***, 0.0259***, 0.0181***), indicating that lowering economic risk can amplify the rent-induced emissions at low emission levels. At higher quantiles, effects turn negative/significant (τ=0.5: −0.00952***; τ=0.7: −0.0355***; τ=0.8: −0.0727***; τ=0.9: −0.129***) indicating that lowering economic risk mitigates rent-induced emissions at high emission levels. • Financial risk interaction (Inrent×Infr, Table 6): Positive/significant at τ=0.2–0.4 (0.0694***, 0.0658***, 0.0277***), implying lower financial risk amplifies rent-induced emissions at low-to-mid quantiles; negative/significant at τ=0.1 and τ=0.5–0.9 (−0.0242***, −0.0270***, −0.0965***, −0.102***, −0.0946***, −0.190***) indicating mitigation at higher quantiles. • Political risk interaction (Inrent×Inpr, Table 7): Consistently negative and significant across τ=0.1–0.9 (−0.107*** to −0.208***), showing that lower political risk (greater stability) weakens the positive effect of resource rents on emissions across the distribution. - Threshold effects (Tables 8–9): Bootstrap linearity tests reject linearity (p=0.000) for all three thresholds; AR(1) p=0.000, AR(2) p=0.180, Hansen p=0.983. • Economic risk threshold: ln threshold γ=3.498 (CI: 3.376–3.620). Below threshold (higher risk): Inrent=0.0259***; above threshold (lower risk): Inrent=−0.0074**. • Financial risk threshold: ln γ=3.214 (CI: 3.060–3.369). Below threshold (higher risk): Inrent=0.0198 (positive, often interpreted as significant in text); above threshold (lower risk): Inrent=0.00949 (not statistically significant). • Political risk threshold: ln γ=3.957 (CI: 3.825–4.090). Below threshold (higher risk): Inrent=0.0273**; above threshold (lower risk): Inrent=−0.0217**. In the conclusions, corresponding index thresholds are reported (approximate ICRG scale): economic risk >36.77, financial risk >35.84, political risk >65.55 associated with mitigation or non-increase of emissions from rents.

The findings demonstrate that dependence on natural resource rents generally elevates carbon emissions, particularly at higher emission levels, consistent with the carbon curse and Dutch disease mechanisms where resource-led industrial structures are carbon intensive. Risk conditions fundamentally shape this nexus. Lower economic and financial risks can enhance financing and innovation capacity. At low-emission stages, this can accelerate economic activity and resource exploitation, amplifying emissions from rents. At higher emission stages, however, improved risk conditions support cleaner technologies, green finance, and policy effectiveness, thereby attenuating the emissions response to greater rents. Political stability exerts a consistently moderating effect: better governance weakens the rent–emissions link across all quantiles and, beyond a threshold, can flip the effect of rents to emissions-reducing—reflecting stronger environmental regulation, effective implementation, and reduced corruption that channels rents towards greener trajectories. The threshold analyses confirm nonlinearities: once economic and political risks fall below certain levels (i.e., stability is higher), increases in rents can be associated with reduced emissions, underscoring the role of institutional and macro-financial quality. Policy relevance: building resilient economic and financial systems (price stability, steady growth, fiscal discipline, exchange rate stability, prudent debt management) and strengthening governance (anti-corruption, rule of law, environmental regulation) are critical to decarbonizing resource-based economies. The heterogeneous quantile effects suggest tailoring policies to countries’ emission stages; in low-emission contexts, risk reduction should be coupled with explicit green investment mandates to avoid rebound in emissions, while in high-emission contexts, improved risk profiles can be leveraged for rapid green transition.

- Natural resources rents are positively associated with carbon emissions across most of the emissions distribution (τ=0.1–0.9), with the 0.8 quantile an exception (insignificant). Robustness checks (POLS, 2SLS, system GMM) confirm this relationship. - Risk moderation: Reduced political risk consistently weakens the positive impact of resource rents on emissions across all quantiles. Economic and financial risks show asymmetric moderation—at low quantiles, lowering these risks can amplify rent-driven emissions; at higher quantiles, lowering risks mitigates them. - Thresholds: Dynamic panel threshold models reveal critical nonlinearity. Above thresholds (approximate index values: economic risk ≈36.77; financial risk ≈35.84; political risk ≈65.55), the typical emissions-raising effect of resource rents diminishes or reverses: economic and political stability can turn rent increases into emission reductions; with financial stability, the rent effect becomes non-significant. Implication: Effective risk management—macroeconomic stability, sound financial systems, and strong governance—can decouple natural resource dependence from rising emissions. Future policies should integrate risk reduction with explicit green finance and regulation to ensure rents support decarbonization. Future research should differentiate among resource types (oil, coal, gas, minerals, forests) to capture heterogeneous production methods and ecological impacts and refine policy design.

- The analysis aggregates diverse natural resources into a single rent measure; effects may differ by resource type (oil, coal, gas, minerals, forests). - While endogeneity is addressed via 2SLS and system GMM and nonlinearity via threshold models, data limitations and index-based risk measures may not capture all institutional nuances. - The study focuses on country-level aggregates across 66 countries; subnational heterogeneity and specific policy interventions are not separately identified. - Risk indices and logarithmic transformations may complicate direct interpretation across scales; conversions between log thresholds and original index levels entail assumptions.