Renewable Energy Consumption and Carbon Emissions: Evidence from an Oil-Rich Economy

Abstract

This article examines the influence of renewable energy consumption, real GDP per capita, exports and imports on consumption-based CO₂ emissions in Azerbaijan from 1993 to 2019 by employing the Dynamic Ordinary Least Squares Method (DOLS). The results reveal that renewable energy consumption has a negative impact on CO₂ emissions, while real GDP per capita has a positive effect. According to the findings, a 1% increase in renewable energy consumption leads to a 0.26% decrease in consumption-based CO₂ emissions, while a 1% rise in real GDP per capita leads to a 0.46% rise in consumption-based CO₂ emissions. In addition, imports and exports show positive and negative effects respectively. Numerically, a 1% rise in imports results in a 0.18% rise in CO₂ emissions, whereas a 1% increase in exports reduces CO₂ emissions by 0.16%. This is consistent with expectations and theoretical outcomes described in the functional specification and data section. The negative influence of renewable energy consumption, as well as the larger effect of imports, emphasize the necessity of implementing ecologically friendly measures in both energy sectors (particularly, the need to increase the share of renewable energy in total energy use) and international trade.

1. Introduction

A significant increase in carbon emissions is one of the main and most challenging issues of the modern global economy [1]. Based on the report of the World Bank, the aggregate level of CO₂ all over the world grew almost 70% from 1990 to 2018 [2]. This is an alarming fact that reflects the rapidly increasing greenhouse emissions. Taking this situation into account, the Intergovernmental Panel on Climate Change (IPCC) reports that global warming could approach 1.5 °C from 2030 to 2052 if the current trend continues [3,4].

Regarding potential environmental effects, greenhouse gas emissions can form several types of pollutions. The most notable one among them is air pollution, which leads to climate change [5]. At present, rising levels of CO₂ in the air lead to global warming and as a result, hazardous effects, such as desertification (especially in Africa), floods (mainly in Indonesia), deforestation, erosion and others emerging in different parts of the world [6]. These natural problems can affect all sectors of the global economy, including health, food security, and many others. Realizing the importance of dealing with CO₂ emission and its consequences, several steps were taken by countries through the Kyoto protocol and the Paris agreement with the aim of decreasing greenhouse gas emissions. In addition, to avoid harmful consequences of conventional energy and CO₂, the United Nations set a goal to shift toward clean and affordable energy (SDG 7). This goal is among 17 important Sustainable Development Goals of the United Nations [7].

Given the significance of developing strategies to minimize greenhouse gas emissions, several researchers are examining this subject in an effort to identify viable solutions to this global issue. Nevertheless, the methods and policy consequences of these findings should be applied so as not to diminish the quality of life in different nations. One of the best ways to reduce CO₂ is to shift toward renewable energy, which is a good substitute for conventional energy. The argument that the shift to renewable energy would lead to a decrease in CO₂ is supported by the fact that the primary renewable energy sources, solar and hydropower, do not create greenhouse gases [8,9,10]. Hence, shifting toward renewable energy will help to reduce hazardous emission levels.

Among the family of nations, resource-rich countries emit more harmful gases and pollute the air more than the other countries [11]. In this regard, finding ways to reduce CO₂ emissions for these resource-abundant nations will be extremely important. Therefore, in this context, we analyzed the effect of CO₂ on renewable energy in Azerbaijan, the biggest oil producing country after Russia in the Caucasian region.

Azerbaijan is one of the most oil-rich nations and was ranked 20th according to its potential oil reserves, while taking 24th place in oil production (barrels per day) [12]. Along with this, it is also rich in renewable energy resources, which makes it a special case for this research and increases the uniqueness of our investigation. Considering the fact that in 2002, aggregate air polluting emissions were 620 thousand tons in Azerbaijan, the level of the same indicator nearly doubled and approached 1122.0 thousand tons in 2019 [13]. Meanwhile, the aggregate air pollutant emissions rose by 80.9% with an average of 4.06% annual growth rate for the time span of 2002–2019 [13,14]. Regarding the economic development of the country, it has demonstrated a remarkable economic growth starting from 2006. For 1996–2019, the GDP of the country grew by almost 29.9 times, from 2733 million AZN in 1996 to 81,681 million AZN in 2019 [15]. However, this economic growth is not without some unpleasant trade-offs and it might lead to significant negative effects on the environment of the nation through different channels. Eventually, unchecked environmental degradation could cause undesirable impacts on individuals and society. Therefore, in the interest of keeping a proper balance among the elements of development, or more clearly, to provide and pursue sustainable development, existing resources of the country should be deployed in an environmentally friendly manner.

Meanwhile, as a result of the harmful influence of conventional energy production and use on the environment, as well as the inefficiency of traditional energy resources, the use of renewable energy sources is becoming increasingly necessary [16]. Therefore, shifting toward clean and renewable energy is one of the most important objectives on the agenda of many countries. International Renewable Energy Agency (IRENA) defined energy transition as a roadmap toward converting the global energy industry into the one in which carbon-free energy resources replace fossil-fuels counterparts, and this process must be done by the second half of the century. Renewable energy, which is also called “clean energy”, can be broadly explained as energy obtained from solar, wave, geothermal, tide, wind, wood, waste, and plant materials (biomass) [17]. The most valuable, attractive and distinctive characteristic of renewable energy is that it can help in three critical areas: pollution reduction, energy security, and long-term economic growth. Regarding higher levels of environmental discharge and the global urgency of the issue, pollution reduction is getting a lot of attention recently. In addition, the International Energy Agency (IEA) sets environment concerns at the center of the renewable energy transition [5]. Moreover, some recent studies such as Ali et al.’s [18], Ding et al.’s [19], Dong et al.’s [20], Aziz et al. [21], and Bhattacharya et al.‘s [22], found that renewable energy lessened environmental degradation.

With the world’s attention and concern focused on global warming and its consequences, studying the effect of renewable energy consumption on CO₂ emissions will be a timely exercise in the case of an oil-rich country, Azerbaijan. Therefore, the aim of this paper is to investigate the relationship between renewable energy consumption and CO₂ emissions in Azerbaijan. For this purpose, the Dynamic Ordinary Least Squares Method (DOLS) was used to test the long run cointegration link between relevant variables.

Azerbaijan has significant potential for the development of clean energy sources; more precisely, the country has tremendous solar (23,000 MW) and wind (3000 MW) resources, as well as important prospects for hydropower (520 MW), biomass and geothermal (380 MW) [23,24]. This gives the country a great opportunity to utilize them and, eventually, shift toward clean energy. However, practical deployment has been limited compared with the scale of the country’s available resources and long-term ambitions. This could be explained by the country’s abundant oil and gas resources, which are less expensive than renewable alternatives. Meanwhile, the gas and oil industries not only impede the transition to renewable energy sources, but they also contribute significantly more CO₂ emissions to the environment than other industries. For instance, in 2021, gas and oil industry accounted for 26.03 million t and 11.11 million CO₂ emissions, respectively. However, cement, flaring, and coal were responsible for 1.1 million t, 252.8 thousand t and 3.6 thousand CO₂ emissions, respectively [25]. For the last two decades, oil and gas accounted for, on average, 98% of CO₂ emissions in the country [2].

This paper contributes to the existing literature in several ways. To begin, to the best of our knowledge, no empirical study has examined the effects of renewable energy consumption on CO₂ emissions in Azerbaijan employing the Dynamic Ordinary Least Squares Method (DOLS). At the same time, Azerbaijan provides a unique opportunity to investigate the link between CO₂ and renewable energy because the country is rich in both natural and renewable energy resources, and it currently aims to have significant investments in renewable energy. Furthermore, we deliberately chose consumption-based CO₂ emissions data in our analysis, which enables us to consider the effect of international trade. This will be very helpful to notice carbon emissions produced in one country and consumed in another [5,26,27]. Another contribution is that we also included other variables, such as exports, imports, and GDP per capita in the model, which were considered in a few previous studies for Azerbaijan. Notably, the model specification is theoretically grounded and proposed by a remarkable study such as Hasanov et al.‘s [5], rather than many studies in the literature that examine the variables of interest in an ad hoc manner. Additionally, we used both imports and exports as a measure of international trade in our research, since combining indicators like trade openness and trade turnover does not enable us to evaluate the distinct effects of exports and imports. As one of the primary drivers of globalization, the increase of international commerce makes it crucial to account for environmental degradation [5].

The remaining part of the paper proceeds as follows. Related studies are reviewed in Section 2. This is followed by econometric specification and estimation in Section 3. Empirical findings are discussed in Section 4, and Section 5 presents conclusions and policy implications.

2. Literature Review

Factors that affect CO₂ emissions and its implications have been extensively studied. Mikayilov et al. [28], Liddle [29,30], Hasanov et al. [31], and Ding et al. [32] investigated the determinants of consumption-based CO₂ emissions. Liddle [29,30] examined the impact of GDP, exports and imports on consumption-based CO₂ emissions and found that GDP and imports have a positive effect on consumption-based CO₂ emissions, while exports have a negative effect on consumption-based CO₂ emissions. In addition, Hasanov et al. [31] found that there is a positive impact from GDP and imports on consumption-based CO₂ emissions, in contrast with a negative impact form exports on consumption-based CO₂ emissions in the case of nine oil-exporting countries. Moreover, Ding et al. [19] revealed a positive effect of GDP and imports on CO₂ emission in contrast with a negative influence of renewable energy consumption and exports on CO₂ emission.

In this paper, we focus on the research which studies the impact of renewable energy consumption (RE), real GDP per capita, exports and imports among other factors, on CO₂ emissions. Zakarya et al. [32] study cointegrates GDP per capita, CO₂ emission per capita, FDI and the total energy consumption using Granger causality in panel. They report a unidirectional causality from CO₂ to other listed variables. Bastola and Sapkota [33] examine causal relationship between real GDP, CO₂ emission and energy consumption using ARDL (Autoregressive Distributed Lag) bounds tests in one of the most climate vulnerable countries, Nepal. They find unidirectional causality from real GDP growth to both CO₂ emissions and energy consumption. Considering Kazakhstan to be an oil-rich post-Soviet country such as Azerbaijan, studies should be interesting. To examine the effect of economic growth, which is proxied by GDP per capita on CO₂ emissions, Akbota and Baek [34] employ autoregressive distributed lag (ARDL) cointegration for the period of 1991–2014. Their findings show that GDP growth increases CO₂ at a low level of income; however, it decreases it at a high level, which means Environmental Kuznets Curve (EKC) holds for CO₂ emissions in Kazakhstan. Hasanov et al. [35] studied the impact of GDP per capita on CO₂ emission per capita from 1992–2013 using various cointegration methods. They report monotonically increasing the impact of GDP on carbon emission. In addition, Mikayilov et al. [27] investigate the impact of international tourism on consumption-based CO₂ emissions adding import and export variables as additional explanatories in Azerbaijan from 1995 to 2013. Employing Fully Modified Ordinary Least Square (FMOLS) method, they find an increase in international tourism raises CO₂ emission. Thio et al. [36] employed STIRPAT model to gauge impact of GDP per capita, technological innovations, exports and imports on CO₂ emissions in 10 selected countries, including Mexico, India, China, South Africa and others. Their findings show that an increase in GDP per capita has significant positive impact on carbon emissions.

In addition, Apergis et al. [37] studied the relationship between CO₂ emission and renewable energy consumption for 19 developed and developing economies utilizing panel data. They revealed that there is a positive effect of renewable energy consumption on CO₂ emissions. Additionally, Khoshnevis Yazdi and Ghorchi Beygi [38] found a negative effect of RE on CO₂ emissions in the case of 25 African economies using Pooled Mean Group (PMG) approach. Dong et al. [20] evaluated the influences from renewable energy consumption on CO₂ emission in the case of 120 countries (including Azerbaijan). The empirical findings indicated that there is a negative and insignificant effect of renewable energy consumption on CO₂ emissions. Moreover, panel cointegration tests were performed by Nguyen and Kakinaka [39] to examine the relationship between renewable energy consumption and carbon emissions for 107 economies throughout the 1990–2013 period. In the case of low-income economies, they found a long-term a positive influence from renewable energy consumption on CO₂ emissions, but a negative effect was seen in the case of high-income economies. Aziz et al. [21] studied the effect of renewable energy consumption on CO₂ emissions in MINT (Mexico, Indonesia, Nigeria, Turkey) countries and found that renewable energy consumption negatively affects CO₂ emissions. Moreover, Mahmood et al. [40] for Pakistan, Leitão et al. [41] for BRICS, Li and Haneklaus [42] for China, Li and Haneklaus [43] for India, and Li and Haneklaus [44] for China reached a negative and significant impact from renewable energy consumption on CO₂ emission. Hasanov et al. [5] analyzed the impact of total factor productivity (TFP) and renewable energy consumption on consumption-based CO₂ emission for BRICS economies using cross-sectional augmented Autoregressive Distributed Lags (CS-ARDL) technique and found negative influence of both the variables on the CO₂ emissions. In addition, the negative effect of renewable energy consumption showed by several studies, such as Bilgili et al.’s [45], Bhattacharya et al.‘s [22], Khoshnevis Yazdi and Shakouri’s [46], Dogan and Seker’s [47], Zoundi’s [48], Danish et al.’s [49], Waheed et al.’s [50], Adams and Acheampong’s [51], Cheng et al.’s [52], Akram et al.’s [53], Saidi and Omri’s [11], Piłatowska et al.’s [54], Leitão and Lorente’s [55], Shahnazi and Dehghan Shabani’s [56], and Ali and Kirikkaleli’s [18].

The literature study reveals that no study has been undertaken on the influence of renewable energy consumption on consumption-based CO₂ emissions in the case of Azerbaijan utilizing time-series analysis. Hence, the purpose of this research is to fill this void by examining the long-run consumption-based CO₂ emissions impact of the renewable energy consumption, alongside real GDP per capita, imports and exports, using DOLS method.

3. Econometric Methodology and Data

3.1. Functional Specification and Data

For BRICS countries, Hasanov et al. [5] proposed a framework in which the consumption-based CO₂ emissions is a function of renewable energy consumption, total factor productivity (TFP), income, imports and exports. In this article, TFP data has not been included in the model specification because it is not available for Azerbaijan. Following the econometric specification suggested by Hasanov et al. [5], the consumption-based CO₂ emissions is modelled as a function of renewable energy consumption, real GDP per capita, exports and imports as follow:

$$lnC{O}_{2,t}={\beta }_0+{\beta }_{1}lnR{E}_t+{\beta }_{2}ln{Y}_t+{\beta }_{3}lnEx{p}_t+{\beta }_{4}lnIm{p}_t+{\epsilon }_t$$

(1)

where, CO_2,t represent consumption-based carbon dioxide emissions, RE_t represents renewable energy consumption Y_t represents gross domestic product per capita, Exp_t represents exports of goods and services, Imp_t represents imports of goods and services and ε_t is an error term. β₁, β₂, β₃, and β₄ indicate the elasticities of consumption-based carbon dioxide emissions with respect to renewable energy consumption, real GDP per capita, export and import, respectively.

Consumption-based carbon dioxide (CO₂) emissions is the dependent variable. It is expressed in million tons of CO₂, (MtCO2). According to recent research, considering consumption-based CO₂ rather than territorial-based CO₂ is preferable [5,27,29,30,31]. One of the benefits of using consumption-based CO₂ as a proxy of carbon emissions is that it includes emissions from both final consumption and international purchasing [5,26]. It has been adjusted to account for international trade and so makes it simple to detect carbon emissions produced in one country and consumed in another [57]. RE is renewable energy consumption as percentage of total energy use. Consumption of renewable energy is derived from non-fossil fuel sources. It is predicted that RE will reduce consumption-based CO₂ emissions. GDP per capita (Y) is proxied in US dollars at 2010 prices. This is the market value of all final goods and services produced in Azerbaijan during a period of time, expected to have a positive effect on consumption-based CO₂ emissions. Exports (Exp) are proxied at constant 2010 US dollars as percentage of GDP. Imports (Imp) are proxied at constant 2010 US dollars as percentage of GDP. Based on the theoretical approach suggested by Hasanov et al. [5], we can examine the CO₂ impact of exports and imports separately instead of combining them. Exports are predicted to have a negative influence on consumption-based CO₂ due to the measure of goods and services produced in one economy but consumed in another. Imports are predicted to have a positive impact on consumption-based CO₂ due to measures of goods and services produced in abroad but consumed in the domestic country [5,27,29,30,31].

The data of RE, Y, Exp and Imp are compiled from World Bank database [2]. The data for CO₂ emissions are compiled from the Global Carbon Atlas [58]. The logarithmic form is used for all variables. The annual data from 1993 to 2019 are utilized.

3.2. Econometric Methodology

The variables are initially checked for properties of stationarity. The Augmented Dickey Fuller (ADF) [59] and Phillips–Perron (PP) [60] unit root tests are employed for this. In the second stage, the presence of long-run co-movement between the variables is checked utilizing Engle–Granger [61], Phillips–Ouliaris [62], and Park’s Added Variables [63]. The Dynamic Ordinary Least Squares Method (DOLS) [64] is used to assess the long-run influence of RE, Y, Exp and Imp on CO₂ emissions. The Dynamic Ordinary Least Squares (DOLS) method was proposed by Stock and Watson [64] as an asymptotic efficient estimation tool. In this approach, an evaluation methodology is proposed that eliminates the endogeneity problem in the cointegration system. DOLS is used to estimate the equilibrium corrected for potential simultaneity bias between variables in the long run [64]. For the cointegration analysis, the OLS estimates are super consistent, as shown by Engle and Granger [61]. They, however, suffer from simultaneity bias issue caused by endogeneity, and standard inferential statistics (based on the t and F statistics) are invalid. The DOLS method, like its counterparts FMOLS and CCR, is designed to overcome these issues. In addition, DOLS, unlike FMOLS and CCR, provides a dynamic framework. Furthermore, Stock and Watson [64] show that this method is more favorable, particularly in small samples, compared to a number of alternative estimators of long-run parameters, including those proposed by Engle and Granger [61], Johansen [65] and Phillips and Hansen [66]. DOLS, like any other methods, has limitations as well. For example, Pesaran and Shin [67] show that their ARDL method is better than DOLS in terms of having less bias and inconsistent estimates. Moreover, DOLS, like any other single equation or residual based methods, assumes one or no cointegration. However, the theory of cointegration articulates that n variables can establish n-1 cointegration at maximum (e.g., see Engle and Granger).

We do not present details about the aforementioned methods. Because we do not want to bore the readers with complex mathematical tools that are widely used in many studies. Dickey and Fuller [59], Phillips and Perron [60], Engle and Granger [61], Phillips and Ouliaris [62], Park [63], and Stock and Watson [64], among others, present comprehensive information.

4. Empirical Results and Discussion

The stationarity properties of the utilized variables are first evaluated using the ADF and PP unit root tests.

Table 1. Findings of unit root tests.

Variable	The ADF Test				The PP Test
	Level	k	First Difference	k	Level	First Difference
CO2	−1.645	0	−10.461 ***	0	−1.351	−9.730 ***
RE	−2.415	0	−5.764 ***	0	−2.399	−5.848 ***
Y	−1.685	1	−2.746 *	1	−0.427	−2.953 *
Exp	−0.787	1	−9.030 ***	0	−1.041	−3.097 **
Imp	−2.010	1	−3.352 **	1	−1.458	−2.716 *

Notes: ADF and PP, respectively, stand for Augmented Dickey–Fuller and Phillips–Perron tests; ***, **, and *, respectively, stand for null hypothesis rejection at 1%, 5%, and 10% significant levels.

shows the results of the unit root tests. According to tests results, all variables become non-stationary at their levels but stationary at the first difference. As a consequence, the cointegration link between the utilized variables may be examined.

Then, the Engle–Granger, Park’s Added Variables, and Phillips–Ouliaris cointegration tests are used to determine the cointegration relationship, and the findings are provided in

Table 2. Outcomes of cointegration tests.

Park’s Added Variables Test		Phillips–Ouliaris
Chi-square	4.45609 (0.21)	Tau-statistics	−5.4113 (0.00)
		Z-statistics	−31.851 (0.00)
Engle–Granger
Tau-statistics	−5.2978 (0.01)	Z-statistics	−27.916 (0.01)

Notes:p-values are in parenthesis.

.

The critical values of the Z-statistics and Tau-statistic of the Engle–Granger and Phillips–Ouliaris tests reported in the Table 2 are greater than the corresponding critical values, meaning that the null hypothesis of “no cointegration” can be rejected in favor of alternative hypothesis of cointegration. This is the opposite for the Park’s added variable test, where the sample reported in the Table 2 is lower than the corresponding critical value, meaning that the null hypothesis of cointegration cannot be rejected in favor of alternative hypothesis of “no cointegration”. Therefore, the results of the tests show that the variables have a long-run cointegration relationship.

As a result, using DOLS technique, after validating the existence of cointegration relationships between the variables, the long-term impact of RE, Y, Exp, and Imp on CO₂ emissions may be assessed.

Table 3. The results of DOLS.

Regressor	Coefficients	t-Statistics	Probabilities
RE	−0.26	−2.3478	0.04
Y	0.46	36.901	0.00
Exp	−0.16	−2.0733	0.06
Imp	0.18	3.6929	0.00

Note: CO₂ is dependent variable.

summarizes the findings.

The estimation findings show that RE has a statistically significant and negative influence on consumption-based CO₂ emissions, as shown in Table 3. According to the findings, a 1% rise in RE decreases consumption-based CO₂ emissions by 0.26%. A rise in consumption of renewable energy derived from non-fossil fuel sources will decrease CO₂ emissions. Our results are consistent with outcomes of Bhattacharya et al. [22] for 85 developed and developing countries, Nguyen and Kakinaka [39] for 107 economies, Khoshnevis Yazdi and Ghorchi Beygi [38] for 25 African economies, Dong et al. [20] for 120 countries (including Azerbaijan), Aziz et al. [21] for MINT economies, Mahmood et al. [40] for Pakistan, Leitão et al. [41] for BRICS, Li and Haneklaus [42] for China, Li and Haneklaus [43] for India, Li and Haneklaus [44] for China, and Hasanov et al. [5] for RICS countries. Furthermore, we revealed that real GDP per capita has a positive and statistically significant impact on CO₂ emissions. It implies that a 1% rise in real GDP per capita is accompanied by a 0.46% increase in CO₂ emissions. Our results are in line with economic theory. From a theoretical point of view, an increase in production or income is linked to increased consumption of intermediate and final good and services, resulting in increased CO₂ emissions. From a theoretical standpoint, high economic growth requires more energy as a main input of the production of goods and services. However, the use of traditional energy sources (fossil fuel) to create energy contributes to environmental deterioration by emitting carbon dioxide (CO₂) into the atmosphere, which raises the world average temperature [4]. Since 2006, Azerbaijan’s economy has seen a substantial expansion from an economic perspective. The GDP of Azerbaijan expanded by 10.4 times between 2000 and 2021, from USD 5.27 billion to USD 54.62 billion [68]. Additionally, approximately 98% of all energy used in Azerbaijan comes from fossil fuels [69]. This fact confirms the estimation results of the current study. Moreover, our estimation findings are consistent with the results of previous studies, such as Brizga’s [70], Shuai et al.’s [71], Chaabouni, and Saidi’s [72], Yang et al.’s [73], Mikayilov et al.’s [74], Hasanov et al.’s [31], Chisti and Sinha’s [75], Danish’s [76], Hasanov et al.’s [5] and Mukhtarov et al.’s [77], which found a positive influence of GDP on CO₂ emissions in the case of different developing economies. In addition, export and import effects are revealed to be negative and positive, respectively, which is consistent with expectations and theoretical outcomes described in the theoretical framework section. According to the estimation results, a 1% rise in imports increases CO₂ emissions by 0.18%, while a 1% rise in exports will reduce CO₂ emissions by 0.16%. One explanation for these findings might be that Azerbaijan puts emphasis primarily on the export of raw materials/resources (particularly, crude oil, natural gas, kerosene and so on) and imports most consumption- and production-based goods and services. Thus, Azerbaijan is an oil-rich developing country, and the main portion of its exports consists of crude oil and gas (92% of total exports). Azerbaijan exports CO₂-intensive oil products such as crude oil (including gas condensate), natural gas, motor gasoline, gold, cotton, polymers of ethylene, etc. [78]. In addition, Azerbaijan is a less diversified economy which continues to import the majority of consumption- and production-based products [27]. Azerbaijan mainly imports products such as motor cars and other motor vehicles, petroleum oils (not crude), electrical apparatus, etc., which increase carbon emissions [78]. These facts support the conclusion that exports and imports have a negative and positive effect on CO₂ emissions, respectively. As we saw in Section 3, when any country exports more products, it domestically consumes a fewer goods and services, resulting in a decline in consumption-based CO₂ emissions. On the other hand, more imports imply increased domestic consumption and, as a result, increased consumption-based CO₂. It’s worth mentioning that Hasanov et al. [26] found a positive effect of imports on consumption-based CO₂ emissions, whereas exports have a negative effect for oil-exporting developing countries, Mikayilov et al. [27] for Azerbaijan, Hasanov et al. [5] for BRICS countries, Liddle [29] for 117 countries, Liddle [30] for 20 Asian economies, and Ding et al. [19] for G-7 nations.

5. Conclusions

The study analyzes the effect of renewable energy consumption, real GDP per capita, exports, and imports on consumption-based CO₂ emissions. We tested variables for a unit root as a first step, and the findings confirmed their stationarity in first-differenced form, allowing us to test variables for a common long-run trend. The cointegration link between renewable energy consumption, real GDP per capita, exports, imports and consumption-based CO₂ emissions in Azerbaijan was estimated using the Engle–Granger, Park’s Added Variables, and Phillips–Ouliaris tests. Finally, we performed the DOLS method to evaluate the long-run link among these variables. According to DOLS’s estimation results, both renewable energy consumption and exports reduce CO₂ emissions in the long-term, with a 1% increase in renewable energy consumption and exports resulting in 0.26% and 0.16% fall in CO₂ emissions, respectively. On the other hand, empirical results revealed that GDP per capita and imports have a positive effect on CO₂ emissions. More specifically, 1% increase in GDP per capita and imports will lead to a 0.46% and 0.18% increase in CO₂ emissions, accordingly.

Based on the outcome of the empirical analysis, the negative effect of increasing renewable energy consumption on CO₂ emissions can be explained from the perspective of the substitution effect. This means people will meet some portion of their energy demand through renewable energy. Therefore, the rising usage of clean renewable energy will substitute and drive out some portion of conventional energy. Eventually, lower level of traditional energy will be needed for consumption purposes. As conventional energy resources are main cause of greenhouse gas emissions, decrease in their usage will lead CO₂ emissions to lessen.

The positive impact of GDP per capita on greenhouse gas emissions is plausible from an economic point of view. This is because increasing real production (rise in GDP per capita) can only be achieved by using more energy. Therefore, demand for energy will increase during the time of rising GDP per capita. To meet rising energy demands in this period, the country will try to utilize more affordable energy resources. In the case of Azerbaijan, oil and gas is more affordable than renewable energy sources. Therefore, country prefers the usage of oil and gas reserves to generate energy, which consequently increases CO₂ emissions.

However, initial steps were performed in the direction of increasing clean energy production. For instance, recently, 12 big and 7 small hydroelectric plants were constructed in Azerbaijan. In addition to this, the country built 6 wind, 10 solar, and 6 biomass power plants during 2018–2020 that own an installed capacity of 420 megawatts (MW) [79]. However, to better address the rising energy need, Azerbaijan should efficiently increase investment in renewable energy. If the country invests highly in renewable energy plants using its oil revenues, the cost of generating renewable energy in these plants could be competitive with the cost of producing energy from conventional oil and gas reserves in the near future. Considering all of these, relevant reforms and projects in these directions should be undertaken by the government. In this regard, one plausible reform that should be implemented is slow energy reform. This means eliminating incentives for conventional fuel energy or raising the prices of energy products. Considering the fact that fossil fuel energy consumption accounts for about 98% of total energy consumption [69], it is important to bring up negative incentives, such as slow energy reform. Moreover, according to Mukhtarov [14], higher conventional fuel energy prices will lead to a decrease in CO₂ emissions. From this perspective, this reform could have a positive result on reducing greenhouse gas emissions. On the other hand, if the slow energy reform is applied, the higher revenues will be achieved as a result of increased oil prices. These additional revenues could also be reinvested on new renewable energy projects. Furthermore, low-interest loans, tax exemption and similar government incentives should be utilized to promote clean energy projects.

Consequently, this study has some limitations. Firstly, the data of CO₂ emissions ended in 2020, while the data of renewable energy consumption ended in 2019, so we could not consider the effects of COVID-19 and post-COVID-19 recovery in our research. Our empirical estimation in this study is aggregate, but it would be desirable to carry out a disaggregated analysis of CO₂ emissions for sectors such as hydrocarbon, services, and transportation in order to be capable to suggest sector-specific policies. Lastly, TFP has not been taken into consideration in the analysis due to data unavailability.