Oil-producing countries account for more than 30 percent of global greenhouse gas (GHG) emissions [1
]. Countries such as Russia, Iran, Saudi Arabia, Mexico, and Indonesia belong to the top 20 countries with the largest aggregate carbon dioxide (CO2
) emissions worldwide [2
]. With Saudi Arabia on the top, Russia, Iran, Mexico, Canada, and Brazil also belong to the list of the 20 countries with the largest per capita carbon emissions. The rest of the top carbon-emitting countries, both in cumulative and in per capita terms, are high income and high-middle income economies [3
]. According to Herman Ott, climate scientist and lawyer at ClimateEarth
, states abundant in fossil fuels have been systematically trying to block real action within the framework of the Kyoto Protocol and its successor, the Paris Agreement, by accusing the respective climate change action of heresy [4
]. The potential losses due to the global energy transition due to the consequent reduction of the demand for oil and gas contribute to the obstruction of the deployment of carbon-saving technologies both within these countries and partly abroad over the investment activity of the sovereign wealth funds of the oil-rich countries.
The analysis of the intended nationally determined contributions (INDCs), i.e., the voluntary climate pledges of the oil-producing countries in the framework of the Paris Agreement, shows that oil producers differ in their mitigation strategies from the rest of the world. Their INDCs but also the work of the One Planet Sovereign Wealth Fund Working Group are mostly confined to the increase of renewables’ share in these countries’ energy mixes. By doing so, these countries first and foremost strive for the diversification of the national economies by producing photovoltaic systems and wind turbines. There are no noteworthy pledges of mitigation within the existing manufacturing or energy sectors over the carbon-saving innovations, reduction of the emissions by end consumers, or the carbon tax [1
The literature on the drivers of environmental degradation in fossil-fuel producing and exporting countries is dominated by the parsimonious polynomial ‘black-box’ specifications that test the empirical validity of the environmental Kuznets curve conjecture. The ‘black box’ estimations with the average income as an omnibus variable enable only the assessment of the long-run income-environment relationship (IER). Such estimations allow no inference about the choice of concrete policies to mitigate environmental degradation. Besides, the use of the conventional panel regression techniques that ignore the time series traits of data, and the parametrization of economic specifications, i.e., pre-determinism, could lead to spurious estimators. To address these problems, this study employs the (pooled) mean group and nonparametric panel estimators.
Except Ike et al. [6
], Esmaeli and Abdoullazadeh [7
] and Sadik-Zada and Gatto [8
], there is no study, which addresses the issue of GHGs in oil-producing countries. Nevertheless, all three surveys only assess the carbon footprint of the oil sector and the empirical validity of the environmental Kuznets curve (EKC) conjecture. Despite putting forward a sophisticated theoretical framework in Sadik-Zada and Gatto [8
], their study enables only the compartmentalization of the GHG-footprint on the theoretical level and confines itself to the assessment of the bivariate IER.
In contrast, the paper at hand strives toward disentangling the sources of atmospheric pollution in the oil-abundant settings with a special focus on oil rent as a share of GDP and on sectoral structure, whereby the quantity of the carbon dioxide emissions serves as the proxy of the magnitude of atmospheric pollution. The findings of the study should contribute to the literature on global climate change mitigation. Choosing carbon dioxide as the dependent variable is in line with the notion of the global atmosphere as a sink for global GHGs [9
]. Carbon dioxide is a long-lived climate pollutant and has been recognized in academia as the major driver of global climate change through its greenhouse effects. Sulfur oxides’ or oxides of nitrogen’s environmental repercussions are less severe in terms of global climate change. Methane, a short-lived climate pollutant, which also emanates inter alia from petroleum and especially combustion of the natural gas leads only to the temporary increase of the global temperatures [10
This paper attempts to contribute to the literature on oil abundance and the environment in two ways. First, it represents a structuralist economic theory of natural resources-based development in terms of environmental degradation. The second contribution of this inquiry is empirical. This is the first study, which accounts for experimental evidence on the trade-off between multicollinearity and omitted variable bias. Therefore, it enables the identification of the major drivers of the carbon emissions in oil-producing economies and the reassessment of the EKC hypothesis in the respective context. The study makes use of the novel panel cointegration techniques, which account for the mixed-stationarity, panel heterogeneity, and endogeneity issues. Furthermore, to exclude the model misspecification problem, which is imminent in the parametric specifications, the study also employs novel nonparametric estimators, which are mostly in line with the panel cointegration estimators.
Turning the “black box” of the economy-environment relationship into a “white box” has practical repercussions for the policy formulation in the context of the national climate change mitigation policies in the fossil fuel reliant settings. The contemporary policymaking in the oil-rich settings lacks a reliable empirical basis for the scenario analysis of different environmental policy choices.
The remainder of the manuscript is organized as follows: The literature review in Section 2
analyzes the existing scholarship and puts forward the theoretical framework. Section 3
delves into methodological aspects of the inquiry, puts forward a multi-sector model of environmental degradation, and delineates the theoretical framework. Section 4
is dedicated to data issues. Section 5
presents the empirical estimation results. Section 6
elaborates on the implications of empirical findings and gives policy recommendations.
The data structure is a balanced panel of 37 oil-producing and exporting countries in the time interval spanning between 1989 and 2019. The list of countries is provided in Appendix A.3
. All the variables except for the Political Rights Index (PRI) and Civil Liberties Index (CLI) have been provided by the World Bank, World Development Indicators. PRI and CLI data have been prepared by the Freedom House. The employed data series and their transformations have been outlined in the following Table 1
The study makes use of the elaborations in Törnqvist et al. [92
] and the findings in the follow-up inquiry in Gerdes [93
], which show that in the context of the panel regression analysis, working with the logarithmic share variables yields more efficient estimation results, rather than actual shares when estimating fixed-effects models. Gerdes [93
] shows that the linear estimator inversely weights variations in shares by its denominator.
In the face of the asymmetry of the linear estimator, the implicit weighting of the share variables paves the way for the spurious correlation between the share and the dependent variable and systematic measurement errors [93
]. Hence, to avoid spurious results caused by the combination of shares and fixed effects, the study employs logs of the shares of oil rents, tertiary sector, and fossil fuels as independent variables [96
Furthermore, because the shares of the manufacturing and tertiary sectors’ shares in total GDP are complements, to avoid collinearity the study employs the log-share of the tertiary sector in GDP and the natural logarithm of the absolute value-added of the manufacturing sector as dependent variables.
6. Concluding Remarks
The major aspiration of the present inquiry is the disentanglement of the determinants of the carbon emissions in oil-abundant settings. The survey has been starkly inspired by the results presented in two recent empirical investigations on the greenhouse gas footprint of oil abundance [6
] and their incongruence with the intended nationally determined contributions (INDCs), i.e., the voluntary climate pledges of the oil-producing countries within the framework of the Paris Agreement. The paper at hand proposed a novel conceptual framework that hypothesizes that oil-led deindustrialization and accelerated tertiarization could trigger strong impulses for environmental upgrading. The proposed theoretical framework is in line with the structuralist paradigm of development and environment, which has been dominating both the literature on economic development and the environmental Kuznets curve over four decades. In addition, the study accounts for grave repercussions of oil abundance on the sectoral structure of oil-producing economies.
Following the theoretical framework, the study disentangles the sources of the GHGs in oil-rich settings, whereby the decomposition of gross carbon dioxide emissions by different sectors of the economy, i.e., manufacturing, services, and energy sectors, has been the criterion for the aforementioned compartmentalization. By doing so the study accounts for the sectoral composition theory of the environmental Kuznets curve and scale effects.
The research on the economy-environment relationship is starkly dominated by highly aggregated empirical assessments of the environmental Kuznets conjecture. By focusing on per capita income as an omnibus variable in bivariate specifications, existing studies try to account for multicollinearity and confounding variable problems. Nevertheless, linear and polynomial black box estimations, which dominate the literature on the empirical income-environment relationship, are both problematic because of the high probability of omitted variable bias. Black-box approaches enable only a tendentious bird’s eye perspective on the long-run income–environment relationship. This is why black-box inferences are not sufficient for the formulation of concrete emission mitigation policies.
Based on a more refined methodology, the present paper studies the trade-off between collinearity problems and omitted variable bias on the one hand and the problem of pre-determined econometric parametrization on the other, which allows for concrete inferences about the major economic sectors’ carbon footprints in oil-exporting countries. The results have a high degree of relevance for applied national environmental policies and should be taken into consideration in construing international action on climate change.
The study uses a parametrized pooled mean group, mean group, and dynamic fixed effects estimators, which account for the time series traits of panel data. Besides, the study applies nonparameterized fixed effect panel estimators with varying coefficients. Despite being principally different, both parametric and nonparametric methods give analogous results in terms of the signs and the significance of control variables. The nonparametric estimators complement the pooled mean group, mean group, and dynamic fixed effects estimators by specifying the nonlinear trend between the indicators of individual sectors of the economy, institutional variables, and carbon dioxide emissions.
The study decisively rejects the inverted income-emissions nexus for oil-producing countries and finds strong indications for a positive income-emission relationship. The nonparametric estimator shows that an increasing level of per capita income in the respective time interval has not been associated with a linear but with a rather over-proportionate increase of per capita carbon dioxide emissions. The same holds for the share of oil rent as a percentage of GDP.
The manufacturing sector and the share of fossil fuel-based electricity in the electricity mix also contribute strongly to the carbon footprint of the economy. In contrast to average income and the share of oil rent in GDP, their impact stabilized since the end of the 2000s and the emissions exhibit no more an over proportionate responsiveness to growing GDP-shares of manufacturing and fossils-based electricity generation.
Against the backdrop of the findings related to the significance of fossil fuel-based electricity generation, the pledges of most oil-producing countries regarding the surge of the renewables in the energy mix a is a good step in the right direction.
Further, the estimations show that a growing GDP share of the tertiary sector leads to substantial reductions in carbon dioxide emissions. This vindicates the tertiarization argument behind the environmental Kuznets curve hypothesis. Against the backdrop of the increasing importance of the tertiary sector in growth and employment in developing economies and in economies in transition, the increasing share of a relatively clean services sector gives reason for the hope that economic growth in oil-producing countries will be less carbon-intensive in the future than it has been in the previous three decades. Nevertheless, the study did not differentiate between the increasing share of services in GDP can be attributed to 1. the growth of the tertiary sector, or 2. deindustrialization because of the natural resource curse. This issue should be addressed in the follow-up studies.
Other than in advanced economies, for which literature suggests that democratization leads to greater environmental protection measures, we find that the opposite holds for oil-producing developing economies. In the latter groups of countries, more political rights and civil liberties are not associated with fewer emissions but fewer rights and less freedom lead to reduced emissions. Studies on the development of public transport point in that direction indicating that autocratic elites support environmental upgrading and infrastructure development projects for the sake of improving their international image [105
]. Of course, prior to the operationalization of this empirical finding, further research on this issue is required. Regardless of the motives for better environmental regulations in oil-dependent economies, however, the international community and especially the UN-institutions and international development agencies must take into consideration that oil-reliant autocracies also have a substantial potential for the implementation of environmental protection measures, which could contribute to global climate change mitigation [106
]. Of course, this finding should by no means be misinterpreted as an appreciation for the non-democratic political systems that dominate the oil-rich settings, certainly not, but rather as a plea for more assertiveness in supporting initiatives that enhance climate change awareness and environmental upgrading within this group of countries.