Forest Products Trade-Environment Nexus through the Lens of Carbon Neutrality Targets: The Role of Rural Bioenergy

Abstract

Environmental sustainability is the primary objective of policymakers all around the globe. The most viable option to deal with this situation is to increase the use of renewable energy sources, particularly bioenergy, a carbon-neutral energy source. Trading activities in clean and green products can also enhance environmental performance. The literature on the impact of bioenergy and trade in environmental goods on ecological sustainability is growing. However, the empirical literature has not shed light on the impact of forest products trade (FPT) and rural bioenergy on environmental sustainability, leaving a significant gap in the literature. To address this gap, this analysis examines the impact of FPT and rural bioenergy on environmental sustainability using 23 economies from 2000 to 2022. Empirical estimates of the model are obtained by applying several estimation techniques, such as fixed effects (FE), random effects (RE), two-stage least squares (2SLS), generalized method of moments (GMM), and cross-sectional autoregressive distributed lag (CS-ARDL). The findings confirm that FPT and rural bioenergy reduce CO₂ emissions and contribute to environmental sustainability. The estimates of control variables of economic growth, industrialization, technological development, urbanization, and financial development are positively significant, confirming that these factors increase carbon footprints and thus hurt environmental sustainability. In contrast, political stability negatively impacts carbon emissions and thus promotes environmental sustainability. In light of these findings, policymakers should encourage forest products trade and rural bioenergy to achieve environmental sustainability.

1. Introduction

One of the most vital contributors to climate change and rising global temperature is environmental pollution, which has endangered both human life and ecosystems [1]. In the last few decades, the topic of climate change has become the main point of concern on a global scale. The world’s environmental tale sustainability is largely dependent on controlling climate change. As a result, policymakers and professionals in both developed and developing economies are working hard to find the answer to climate change and global warming issues. In order to overcome these issues, the Paris Agreement 2015 underscores the significance of reaching the highest point of global warming and guarantees that temperature rise remains below 2 °C [2]. Therefore, it is crucial to understand the factors that can impact environmental sustainability to address the issue of climate change effectively and to attain sustainable development goals (SDGs). Thus, a large number of researchers have identified the factors that drive environmental sustainability [3,4]. These factors include but are not limited to globalization, gross domestic product (GDP), energy consumption, trade, information and communication technology (ICT), financial development, foreign direct investment (FDI), etc. However, the role of FPT and bioenergy in impacting environmental quality is unknown, leaving a significant gap in the literature. The primary motive of this analysis is to fill the gap in the literature by analyzing the impact of the FPT, rural bioenergy, and environmental sustainability. The FPT-environment nexus sheds light on the role of forest products as a significant renewable source that is vital in mitigating carbon footprints when managed sustainably. Moreover, the analysis also highlights the significance of bioenergy obtained from trees and crop residue in fostering renewable energy production in the total energy mix of our selected economies. Analyzing the relationship between bioenergy and environmental quality is significant and provides us with important information regarding the role of bioenergy in supporting the goals of the renewable energy transition, carbon neutrality, and rural development. FPT and bioenergy complement each other in achieving carbon neutrality objectives as both are crucial sources of carbon reduction in the ecosystem [5,6]. Thus, the analysis is significant in terms of finding the factors that can contribute to carbon neutrality.

The role of forests is crucial for maintaining a balance in the ecosystem, and their existence is of paramount significance for superior environmental quality. They also play a vital role in absorbing carbon emissions, thereby helping mitigate climate change [7]. Globally, forests absorb nearly 16 billion metric tonnes of carbon dioxide per year [8]. They are also considered one of our best climate solutions. In addition, forests are home to several plant and animal species, hence protecting biodiversity [9]. Recently, the fast depletion of forests has exacerbated the issues of climate change issue. There are several factors that are responsible for the depletion of forests, including agriculture, urbanization, and other socio-economic factors (e.g., poverty, corruption, weak governance). Another crucial factor that has not come under the limelight is the forest products trade [10]. Doha hosted the “Fourth Ministerial Conference of the World Trade Organization (WTO), in November 2001 [11]. The primary motive of this round was to the adoption of “zero-for-zero” tariff policy, which intends to abolish tariffs on all forest products. The Japan rejected the same proposal in the preceding Uruguay Round, prevented from reaching a consensus. In general, a zero tariff policy is for all FPT, including timber trade. The primary motive of this policy is to eliminate import and export duties on FPT as well as timber trade. As a result, this policy enhances the volume of cross-border timber trade by making the timber trade more cost-effective. The zero-for-zero tariff policy can play a significant role in fostering the efficiency of the global timber trade and promoting sustainable timber trade practices [12]. However, policymakers must ensure that trade is made in an eco-friendly manner and based on timber sources that are legally obtained and sustainably managed. To that end, it is crucial to integrate tariff policy with stringent policies that are specifically designed to control the illegal logging and cutting of forests [13].

Since 2001, several theoretical discussions have been carried out with regard to the liberalization of FPT, and there is a consensus that FPT should be liberalized and free from all the hurdles [14]. However, whether the sustainability of the ecosystem, particularly the forest, is linked to the FPT is a research question overlooked in previous discussions. The FPT significantly impacts environmental sustainability as it can disturb the ecological balance and resource availability [15]. Relying on unsustainable procedures during logging and forest product extraction significantly increases deforestation, habitat, and biodiversity loss [16]. As a result, there is a disruption in the ecosystem, leading to a rise in soil erosion, carbon footprints, and climate change. In contrast, FPT can significantly boost environmental quality if properly managed. In this regard, the Forest Stewardship Council (FSC) has provided proper guidelines regarding sustainable forest management. These guidelines state that timber and other forest resources should be harvested without compromising on forest health, which is crucial for fostering biodiversity and regeneration [17,18]. In addition, the Sustainable Forest Initiative (SFI) was introduced by the Program for the Endorsement of Forest Certification (PEFC) in 2005 [19]. The SFI was introduced with the motive to promote sustainable management of FPT by promoting rigorous standards for forest practices. SFI also has a crucial role in maintaining biodiversity, purifying water, and ensuring the conservation of forest resources by certifying forests and forest goods [20]. As far as FPT is confirmed, it assures the traders that the source of the product they trade is legal and the wood is obtained sustainably. This step improves FPT by building the confidence of traders in FPT and increasing their access to certified products [21]. Thus, FPT could impact environmental sustainability in either way, depending on how FPT is managed.

Bioenergy is another factor that is considered by this analysis as a potential determinant of environmental quality [22]. With the growth in renewable energy, the significance of bioenergy is also increasing due to its low-carbon characteristics. Replenishment is one of the main qualities of bioenergy and other renewable energy sources. Since the start of this world, renewable energy sources (e.g., solar, wind, hydel, and biofuels) have been available as the primary sources. Energy generation from these sources results from natural courses that occur repeatedly over our lifetimes [23]. Therefore, renewable energy sources are widely believed to be unlimited sources of energy. Bioenergy is an important source of low-carbon energy generated by firing wood, which has been due to the growing significance of bioenergy, policymakers are now shifting their focus to the increased production of bioenergy as a component of their energy transition strategy [24]. There are three main motives behind their efforts to expand bioenergy production. First, to reduce the reliance on energy sources based on fossil fuel production. Second, a balance between bioenergy production and the chances of lowering carbon emissions can be achieved by limiting coal and lignite consumption. Thirdly, promoting rural development may significantly boost the rural economy [25]. The global CO₂ emissions have risen slightly from 2000 to 2022. The trade of forest products has also increased, while biofuel production has shown steady growth (Figure 1). In 2022, Figure 1 records the highest levels: CO2 emissions at 24.34 tonnes, FPT at USD 20.27 thousand, and biofuel production at 7.08 TWh. Figure 2 shows a gradual increase in the trade values for Asia, Africa, Europe, and America regions. The highest FPT values are recorded in 2022. These are recorded as follows: Asia with USD 19.01 thousand, Africa with USD 16.65 thousand, Europe with USD 19.45 thousand, and America with USD 18.80 thousand. Over the years, biofuel production has significantly increased in America and Europe, with America leading at 749.91 TWh in 2022 (Figure 3).

The literature on environmental sustainability is growing. Previously, the literature has investigated the role of FPT in different contexts such as forest sustainability [26], and forest management [27]. Moreover, studies on renewable energy have considered the role of different forms of renewable energy in environmental sustainability [28,29]; however, the impact of FPT and bioenergy on environmental sustainability has not been analyzed in any past study.

Against this backdrop, the analysis intends to analyze the impact of FPT and rural bioenergy on environmental sustainability. The study is novel to the existing literature in the following aspects. First, as per the known literature, this is the initial effort to investigate the impact of FPT and rural bioenergy on environmental trade. These two factors can impact environmental sustainability in either way; therefore, analyzing their impact on environmental sustainability is an interesting choice and a valuable addition to the existing literature. Second, the findings of the analysis can open the way for future empirical and theoretical analysis in a similar context. Third, the study applies advanced econometric approaches to obtain robust and accurate results: (i) we have applied the preliminary tests, such as Pesaran test for cross-sectional dependence (CSD), Pesaran and Yamagata [30] for slope heterogeneity, and second generation unit root test; (ii) the application of the 2SLS and GMM while empirical estimation of the relationship is helpful in providing robust estimates due to their ability to control endogeneity and individual heterogeneity; (iii) to deal with the non-normal distribution of the data, we employed panel quantile regression; (iv) to determine the short- and long-run impacts of FPT and bioenergy on CO₂ emissions, we employed the CS-ARDL, which controls CSD, endogeneity, and serial correlation. Lastly, the study’s outcomes offer valuable suggestions to policymakers, industry stakeholders, and environmental organizations to achieve environmental sustainability by efficiently utilizing the FPT and rural bioenergy.

2. Literature Review

The study intends to investigate the role of FPT and rural bioenergy in determining environmental sustainability. We have divided the literature into different section. The first section highlights the impact of FPT on environmental degradation, while Section 2 sheds light on the nexus between bioenergy and environmental quality.

The literature on the FPT is very limited. Thus, in Section 1, in order to derive the relationship between FPT and environmental degradation, we have relied on the literature that has estimated the impact of forest resources and trade on environmental degradation. Koirala and Mysami [31] analyzed the connection between forest resources and carbon footprints using ordinary least squares (OLS) in the USA and confirmed that forest resources help foster environmental quality by enhancing carbon footprints. A similar type of relationship was observed by Ahmad et al. [32] in the context of Pakistan. De Sy et al. [33] revealed that land-use change is a major factor that enhances South America’s carbon footprints. Their findings also suggest that remote sensing time series can systematically identify the causes of deforestation and carbon losses in the South American area. Hewson et al. [34] revealed that efficient forest management and preservation of forest resources are vital in mitigating the carbon footprints in the eastern Madagascar region. Using the multiple linear regression model for cross-section data from 61 countries in 2010, Tian et al. [35] noted that trade can enhance the allocation efficiency of timber resources worldwide, improving utilization efficiency and reducing global wood consumption to protect forest resources. Using the difference-in-differences approach, Cary [36] indicated that the competitive advantage in the international timber trade network has been balanced, contributing to the advancement of SDGs focused on economic and environmental objectives.

In the context of China, Nasrullah et al. [37] analyzed the factors that contribute to its trade in forest products. The study gathered data on different forest product groups from 2001 to 2018 and applied the gravity model of trade. Their findings suggest that GDP has a crucial role in fostering trade, while distance hinders the FPT. Olmos (2022) examined the FPT in the context of 19 South and Central American nations and Mexico from 1990 to 2020 and their findings confirm the positive connection between the level of income and FPT. Muhammad and Jones [38] investigated the nexus between tariff exclusion between FPT of the US within the framework of a US–China trade war. According to the generalized Gauss–Newton method outcomes, the study revealed that the US–China trade war hurt FPT and the exclusion of the tariff from China benefited the US. Darrobers et al. [39] scrutinized the US timber trade with its major trading partner from 1870 to 2017 using economy-wide material flow accounting (EW-MFA). They observed that the US during this period relied more on timber imports from Canada, Asia, and Europe in primary and final products, and it emerged as the exporter of raw timber products. By utilizing OLS and random effects (RE) estimation approaches, Magezi and Okan [40] estimated the FPT between the Turkish and European Union (EU) economies. As per their outcomes, the GDP and population positively influence the forest goods exports from Turkey to the EU economies, while the relative forest endowment hurts the FPT.

As far as the nexus between trade and environmental quality is concerned, it could be positive or negative. Empirical works of Chen et al. [41] and Du et al. [42] observed the positive connection between trade and carbon emissions, while Shahbaz et al. [43] used the FMOLS method and observed an inverse link between trade and carbon emissions. On the other, there is some evidence confirming the insignificant impact of trade on environmental degradation [44].

In order to control environmental degradation several empirics have analyzed the role of biomass energy by using different economic measures. However, consensus has not yet been reached with regard to the positive or negative role of biomass in environmental degradation. In the context of BRICS, Wang [45] analyzed the effect of biomass energy on CO₂ emissions by applying GMM from 1992 to 2013. Similarly, Ulucak [46] by relying on the dynamic autoregressive distributed lag (ARDL) revealed the unfavorable influence of bioenergy on carbon footprints from 1982 to 2017. Shahbaz et al. [47] applied the GMM model to the data from 1990 to 2015 in the MENA nations and found the favorable role of bioenergy in fostering environmental quality. Sulaiman et al. [48] in their analysis of 27 EU nations from 1990 to 2017. The results of the dynamic ordinary least square (DOLS) revealed an inverse correlation between carbon emissions and bioenergy. Research conducted by Pathak and Das [49] confirmed that bioenergy has significant environmental, social, and economic benefits. In contrast, the empirical effort by Solarin et al. [50] in the context of 80 developed and emerging nations by assembling the data from 1980 to 2012. They relied on the GMM technique and found the correlation between carbon footprint and bioenergy favorable. For the same period but for 13 Asian nations, Gao and Zhang [51] employed a fully modified ordinary least square (FMOLS) method and revealed that bioenergy fosters environmental quality.

While there have been several studies on CO₂ emissions, there has been little research on the interplay between bioenergy and ecological footprint (ECF). The research conducted by Wang et al. [52] aimed to investigate the relationship using DSUR between bioenergy and ecological footprint in the G-7 countries from 1980 to 2016 and confirmed a favorable correlation between bioenergy and ECF. In contrast, Hadj’s [53] research found that the utilization of biomass energy decreased the ecological footprint in Saudi Arabia from 1984 to 2017, as determined by the use of linear and nonlinear ARDL methodologies.

3. Theoretical Framework

Three effects highlight the theoretical link between trade and environmental quality, such as the scale, composition, and technology effects [54]. The scale effects highlight the rise in the consumption and production activities in the economy [55]. The composition effect highlights the modifications in the basket of finished products, whereas the technology effect leads to changes in the manufacturing technique, particularly the transition towards clean technologies. It is widely believed that the scale effect increases the pressure on the ecosystem, resulting in environmental degradation; however, the technology effect is believed to be environmentally friendly. The impact of composition effects on the environment largely relies on the nature of “comparative advantage” [56]. Thus, as per the notion of comparative advantage, if a nation has an edge over other nations in the production of goods that are not eco-friendly, the composition effect will result in adverse environmental impacts on the nation. On the other hand, if the composition effect helps the country transform its basket of finished goods from polluted to green ones, then the composition effect results in a cleaner environment. In general, the increase in trade may lead to a clean environment if the technology effect overpowers the other two effects, i.e., scale and composition (a country has an edge in producing dirty and polluted items). The positive outcome of trade also appears if the technology and the competition effect (a country has an edge in producing clean items) overpower the scale effect [57]. In simple words, FPT can foster environmental quality if it is conducted in line with sustainable trade practices by using technology and promoting healthy competition. However, FPT can hurt environmental quality if not properly regulated, leading to deforestation and enhanced carbon output in the ecosystem.

In order to enhance economic output, energy is widely believed to be a crucial factor because it is an important part of the production function of every industry [58]. Most of the energy is obtained from fossil-fuel-based energy sources, which are the main contributors to global carbon emissions; thus, heightened energy consumption adversely affects environmental quality. As a result, there is a consensus that increased energy demand is primarily responsible for the degradation of the ecosystem. Thus, reducing energy consumption, mitigating emissions, and decreasing reliance on fossil-fuel-based energy sources and their efficient consumption is one of the available options for policymakers [59]. Enhancing renewable energy generation plays a vital role in achieving the above-stated objectives. Biomass is a widely accessible renewable energy source that can replace fossil fuels in the energy mix and foster environmental sustainability. On the basis of the above discussions, we have developed the following two hypotheses:

H1. 

Forest product trade may increase environmental sustainability.

H2. 

Bioenergy has the potential to improve environmental sustainability.

3.1. Econometric Model

This study aims to scrutinize the impact of forest products trade and rural bioenergy on carbon emissions. Following Verburg et al. [60] and Dogan and Inglesi-Lotz [61] we have constructed the following model and modified it according to our needs:

$${\mathrm{C}\mathrm{O}}_{2,\mathrm{i}\mathrm{t}}={\mathsf{\kappa }}_0+{\mathsf{\kappa }}_1{\mathrm{F}\mathrm{P}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_2{\mathrm{R}\mathrm{B}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_3{\mathrm{E}\mathrm{G}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_4{\mathrm{G}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_5{\mathrm{I}\mathrm{N}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_6{\mathrm{T}\mathrm{E}\mathrm{C}\mathrm{H}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_7{\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_8{\mathrm{U}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_9{\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(1)

where CO₂ emissions (CO₂) are determined by forest products trade (FPT), rural bioenergy (RBE), economic growth (EG), government spending (GS), industrialization (IND), technology (TECH), financial development (FD), urbanization (UR), political stability (PS), and error term (ε_it). FPT can exert a mixed impact on CO₂ emissions. On the positive side, if FPT is managed in ways that maintain ecological processes, productivity, and biodiversity, it will ultimately reduce CO₂ emissions. Moreover, improved forest activities driven by international trade can promote better forest conservation practices, thereby reducing CO₂ emissions. On the negative side, FPT also drives forest degradation and deforestation, particularly when forest logging is not properly managed, resulting in increased CO₂ emissions. RBE reduces dependence on fossil fuels. By substituting fossil fuel energy sources with bioenergy, the CO₂ emissions of energy production can be lowered.

3.2. Econometric Methodology

3.2.1. CSD and Homogeneity Tests

CSD significantly affects panel data analysis. This issue may arise from the interconnections between economies, such as trade integration, shared borders, economic and financial dependencies, and mutual social traits [62]. Because of these connections, a shock in one country can impact all other countries in the region, or even in other regions. This can lead to inaccurate results. Therefore, to obtain accurate estimates, it is essential to check for CSD [63]. The following equation represents the CSD equation:

$$CS{D}_{test}=\sqrt{{\displaystyle \frac{2T}{N(N-1)}}}\left({\displaystyle \sum _{i=1}^{N-1}{\displaystyle \sum _{k=i+1}^N{\widehat{\tau }}_{ik}}}\right)$$

(2)

This study also employed the Friedman test to detect CSD. Another important test is slope heterogeneity, which is essential for obtaining accurate estimates. Accordingly, the slope heterogeneity test developed by Pesaran and Yamagata [30] is utilized, and the equations representing this test are provided below:

$${\tilde{\Delta }}_{HPY}={\left(N\right)}^{{\displaystyle \frac{1}{2}}}{\left(2k\right)}^{-{\displaystyle \frac{1}{2}}}\left({\displaystyle \frac{1}{N}}\tilde{S}-k\right)$$

(3)

$${\tilde{\Delta }}_{ASH}={\left(N\right)}^{{\displaystyle \frac{1}{2}}}\left(2k{\left({\displaystyle \frac{T-k-1}{T+1}}\right)}^{-{\displaystyle \frac{1}{2}}}\left({\displaystyle \frac{1}{N}}\tilde{S}-2k\right)\right)$$

(4)

3.2.2. Unit Root Tests

The CSD and slope heterogeneity tests also assist in determining whether to use first-generation or second-generation unit root tests. If CSD is present in the data, first-generation unit root tests become ineffective and fail to yield accurate results. To address the issues caused by CSD, the second-generation unit root test “cross-sectionally augmented Im, Pesaran and Shin (CIPS)”, extended by Pesaran [64], can be implemented. This test effectively manages both CSD and heterogeneity. The unit root test employed the previously mentioned equations:

$$\begin{array}{ll}\mathsf{\Delta }{\mathrm{V}}_{\mathrm{i},t}=& {\alpha }_{\mathrm{i}}+{\alpha }_{\mathrm{i}}{\mathrm{X}}_{\mathrm{i},t-1}+{\alpha }_{\mathrm{i}}{\mathrm{Y}}_{\mathrm{t}-1}+{\sum }_{\mathrm{l}-0}^{\mathrm{p}}{\alpha }_{i1}\mathsf{\Delta }\overline{{\mathrm{V}}_{\mathrm{t}-1}}+{\sum }_{\mathrm{i}-1}^{\mathrm{p}}{\alpha }_{\mathrm{i}}\mathsf{\Delta }{\mathrm{Y}}_{\mathrm{i},t-1}+{\mu }_{it}\\ & \end{array}$$

(5)

Equation (5) illustrates the cross-section averages denoted by $\overline{{\mathrm{V}}_{\mathrm{t}-1}}$. This equation can give rise to the cross-sectional augmented Dickey–Fuller (CADF) approach, and the CADF value may reach the level of CIPS.

$$\widehat{CIPS}={\displaystyle \frac{1}{N}}{\sum }_{i=1}^n{C}_{ADF}^i$$

(6)

3.2.3. Cointegration Test

Several testing procedures are available for investigation, including those developed by Kao [65], Pedroni [66], and Westerlund [67]. However, the Pedroni and Westerlund tests only allow for the inclusion of seven covariates in it. In this study, we use the Kao cointegration test to examine the relationship between the variables of interest. All tests are based on the estimated residuals from the following long-run model.

$$y_{it}=x_{it}^{\prime }{\beta }_i+z_{it}^{\prime }{\gamma }_i+e_{it}$$

(7)

For each panel, each covariate in x_it is an I(1) series. All tests assume that the covariates are not cointegrated. The Kao, Pedroni, and Westerlund tests employ different approaches to assess whether e_it is nonstationary. The Dickey–Fuller (DF) test statistics are calculated using OLS on Equation (7), predicting residuals ($\widehat{e_{it}}$), and fitting the DF regression model.

$${\widehat{e}}_{it}=\rho {\widehat{e}}_{i,t-1}+{\nu }_{it}$$

(8)

In this context, ρ represents the autoregressive (AR) parameter and ν_it denotes the error term. The DF and unadjusted DF tests evaluate whether the coefficient ρ equals 1. In contrast, the modified DF and unadjusted modified DF tests investigate whether ρ – 1 = 0.

3.2.4. Fixed Effects, Random Effects, 2SLS, GMM, and Panel Quantile Regression

The panel data modeling framework is used in this analysis to conduct empirical analysis. Here are the known advantages of panel data modeling. First, it has the potential to estimate the model more accurately because there is more flexibility and variability in the model. Second, since panel data includes time and space dimensions, it also helps account for complicated human behavior. Third, it can estimate the model well even in the presence of missing observations. The difference between panel data and conventional data requires the construction of special techniques to deal with panel data so that all its benefits can be enjoyed. The simplest of all the panel data techniques is the pooled OLS [68]. The other approaches, FE and RE are also used for estimating panel data, but they have distinct drawbacks. The decision to apply FE or RE is quite simple and based on how we chose the cross-sections. For instance, if cross-sections are chosen at random, RE should be used; however, if we are sure about the cross-sections, FE is a better option. The main limitation of the FE is that it cannot estimate the model that has time-invariant variables, while the limitation of the RE is that it has to rely on the assumption that no correlation between regressors and country-fixed effects, which, if violated, makes the estimate biased. Moreover, the FE and RE overlook the common panel problems such as “endogeneity, serial correlation, and heteroskedasticity” [69].

Furthermore, it disregards individual heterogeneity and endogeneity, two major problems with panel data that might lead to incorrect conclusions [70]. In this situation, instrument variable (IV) approaches, such as “2SLS” and “GMM”, can serve the purpose well. 2SLS is a renowned IV technique of Cumby et al. [71] that can regulate individual heterogeneity and endogeneity among the regressors. This approach estimates the panel model in two steps. In the first step, all exogenous regressors are regressed on the endogenous regressors. In the second, these estimated values are used as instruments in place of the endogenous regressors in the original model to control endogeneity. In 2SLS, it is assumed that the instruments are uncorrelated with the error term and are chosen based on theoretical knowledge. 2SLS is quite effective in resolving endogeneity, but choosing the right instrument is not always simple [72].

$${\mathrm{C}\mathrm{O}}_{2,\mathrm{i}\mathrm{t}}={\mathsf{\kappa }}_0+{\mathsf{\kappa }}_1{\mathrm{F}\mathrm{P}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_2{\mathrm{R}\mathrm{B}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_3{\mathrm{E}\mathrm{G}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_4{\mathrm{G}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_5{\mathrm{I}\mathrm{N}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_6{\mathrm{T}\mathrm{E}\mathrm{C}\mathrm{H}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_7{\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_8{\mathrm{U}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_9{\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\upsilon }}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(9)

Another well-known IV technique is GMM of Arellano and Bover [73] and Blundell and Bond [74], which is easy to apply as compared to other IV techniques. One of the most significant troubles in the application of the IV approaches is how to decide about the appropriate instruments, as comprehensive theoretical and empirical knowledge is required to do so [75]. The GMM addresses this issue by allowing the use of all the right-hand side regressors and the lag of the CO₂ variable as instruments. Increasing the number of instruments reduces the probability of producing wrong estimates as a result of endogeneity [76]. It makes sense to think that the past CO₂ is a driver of the present CO₂; therefore, using lag CO₂ on the right-hand side is appropriate. This can help GMM to deal with the issues of autocorrelation and endogeneity. Consequently, the following model (10) serves as a representation of the GMM’s baseline equation:

$${\mathrm{C}\mathrm{O}}_{2,\mathrm{i}\mathrm{t}}={\mathsf{\kappa }}_0+{\mathsf{\lambda }}_1{\mathrm{C}\mathrm{O}2}_{\mathrm{i}\mathrm{t}-1}+{\mathsf{\kappa }}_1{\mathrm{F}\mathrm{P}\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_2{\mathrm{R}\mathrm{B}\mathrm{E}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_3{\mathrm{E}\mathrm{G}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_4{\mathrm{G}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_5{\mathrm{I}\mathrm{N}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_6{\mathrm{T}\mathrm{E}\mathrm{C}\mathrm{H}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_7{\mathrm{F}\mathrm{D}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_8{\mathrm{U}\mathrm{R}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\kappa }}_9{\mathrm{P}\mathrm{S}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\upsilon }}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(10)

As our data have large (N) and small (T), GMM is best suited for this type of data. GMM comes in two types: difference GMM and system GMM. The difference GMM of Arellano and Bond [77] is the fundamental approach that may suffer from the issue of weak instruments if the data period is short. To counter this situation, system GMM is an appropriate approach that has the ability to deal with short and unbalanced data. The GMM also uses the Hansen test and the AR(2) test to validate the instruments and examine serial correlation.

In a panel quantile regression, it is suggested by Koenker and Bassett [78] to analyze the link between the outcome variable and the regressors to use the conditional quantile functions for the outcome variable contingent on the regressors. The main justification for this approach is the existence of “non-normal data distributions” and using approaches that rely on the conditional mean model may provide erroneous estimates. As a result, it is vital to let the slopes vary for different quantiles of the outcome variable (CO₂ in this case). The environment and energy-related literature is filled with studies that confirm these data sets exhibit outliers Wei and Ullah [79]. This approach also efficiently handles this issue. Thus, it is an ideal approach for estimating the energy and environment data.

3.2.5. CS-ARDL and PMG-ARDL

The robustness estimator in this analysis is the CS-ARDL model. This method was created by Chudik and Pesaran [80] and surpasses its counterparts in terms of benefits. For instance, the CS-ARDL does not make it mandatory that all variables must be of the same order of integration and it can estimate the model with the variables that are based on mixed order of integration, i.e., I(0) and I(1), while all other panel cointegration procedures are only relevant if the variables have a unit root or integrated order one. Moreover, it is among the few panel cointegration approaches that are capable of providing the short- and long-run effects. The ability to hand the “CSD, endogeneity, slope heterogeneity, and serial correlation” during panel analysis makes the CS-ARDL superior to competing techniques. We create the following Equation (7) in order to test the variables in the CS-ARDL framework empirically:

$$\mathsf{\Delta }{CO}_{2, it}={\varnothing }_i+{\sum }_{l=1}^p{\theta }_{il}\mathsf{\Delta }{CO}_{2, it-l}+{\sum }_{l=0}^p{\theta }_{il}^{\mathrm{\prime }}{X}_{it-l}+{\sum }_{l=0}^1{\theta }_{il}^{\mathrm{\prime }}{\overline{W}}_{it-l}+{\epsilon }_{it}$$

(11)

where ${\overline{W}}_t=\left(\overline{{CO}_{2,t}} . \overline{Y_t}\right)\mathrm{and} X_{it}={\left({FPT}_{it}+{RBE}_{it}+{EG}_{it}+{GS}_{it}+{IND}_{it}+{TECH}_{it}+{FD}_{it}+{UR}_{it}+{PS}_{it}\right)}^{\mathrm{\prime }}, \mathrm{and} X$ is the vector of independent variables. To enhance robustness, we used the pooled mean group-ARDL (PMG-ARDL) method [81]. This approach mitigates the autocorrelation problem and simultaneously provides short- and long-term effects. Additionally, the study also employs the panel quantile regression method. This method accounts for the entire distribution while addressing time-varying issues of heterogeneity and outliers. The analysis flowchart is shown in Figure 4.

4. Data and Descriptive Analysis

This study determines the nexus between forest products trade, rural bioenergy, and environmental sustainability using 23 economies from 2000 to 2022. Sample economies are chosen based on data availability (see

Table A1. List of sample countries.

Canada	Finland	UK
Mexico	Germany	Australia
USA	Italy	China
Argentina	Netherlands	India
Brazil	Poland	Indonesia
Colombia	Potugal	Thailand
Austria	Spain	France
Belgium	Sweden

). Environmental quality is proxied by CO₂ emissions, measured in kilotons. The World Development Indicator (WDI) is the source of data for CO₂ emissions. The focused variables are forest products trade (FPT) and rural bioenergy (RBE). Forest products trade (FPT) is proxied by forest product trade, measured in USD 1000. Rural bioenergy (RBE) is proxied by biodiesel production, measured in petajoules. Data sources for forest products trade (FPT) and rural bioenergy (RBE) are the Food and Agriculture Organization (FAO) and British Petroleum (BP), respectively. Economic growth (EG), government spending (GS), industrialization (IND), technology development (TECH), financial development (FD), urbanization (UR), and political stability (PS) are included as control variables in this research to determine how much these parameters contribute to CO₂ emissions.

Studies document both positive and negative effects of economic growth on CO₂ emissions. Balsalobre-Lorente et al. [82] stated that higher economic growth provides the financial resources required to invest in green technologies, potentially reducing CO₂ emissions. Economic growth (EG) is proxied by GDP per capita growth, measured in annual percent. Ullah et al. [83] illustrated that government expenditures in environmental protection, eco-friendly projects, research and development, and environmental infrastructure significantly reduce CO₂. In our study, government spending (GS) is proxied by the general government’s final consumption expenditure, which is measured as a percentage of GDP. Liu and Bae [84] documented a positive nexus between industrialization and CO₂. The study argues that industrialization leads to resource depletion and augments pollution emissions, negatively influencing environmental quality. Industrialization (IND) is proxied by industrial value added, including construction, which is measured as a percentage of GDP. Ullah et al. [85] revealed the importance of technological development for environmental quality. They claim that green technological development and renewable energy can significantly reduce environmental degradation. The technology development (TECH) variable is proxied residents and non-residents total patent applicants. Shahbaz et al. [86] stated that financial development provides funds for green technologies and eco-friendly projects that facilitate environmental quality. Financial development (FD) is proxied by domestic credit provided by banks to the private sector as a percent of GDP. Liu and Bae [84] demonstrated a positive association between urbanization and CO₂. They stated that urbanization leads to increased resource consumption and pollution due to high industrial activities and population densities. Urbanization (UR) is proxied by urban population as a percent of the total population. Ayhan et al. [87] reveal that political stability (PS) allows for effective and consistent implementation of environmental regulations and policies, thus leading to reduced CO₂. This variable is measured through political stability estimates and the absence of violence/terrorism estimates. WDI provides data for EG, GS, IND, TECH, FD, and UR, while the Worldwide Governance Indicator (WGI) provides data for PS. See

Table 1. Variables specification.

Variables	Symbol	Definitions	Sources
CO2 emissions	CO2	CO2 emissions (kt)	WDI
Forest products trade	FPT	Forest products trade (USD 1000)	FAO
Rural bioenergy	RBE	Biodiesel production (Petajoules)	BP
Economic growth	EG	GDP per capita growth (annual %)	WDI
Government spending	GS	General government final consumption expenditure (% of GDP)	WDI
Industrialization	IND	Industry (including construction), value added (% of GDP)	WDI
Technology development	TECH	Patent applications, total	WDI
Financial development	FD	Domestic credit to private sector by banks (% of GDP)	WDI
Urbanization	UR	Urban population (% of total population)	WDI
Political stability	PS	Political stability and absence of violence/terrorism: Estimate	WGI

for a detailed description of data.

Descriptive statistics are reported in

Table 2. Descriptive statistics.

	Mean	Median	Max	Min	SD	Skewness	Kurtosis
CO2	12.60	12.65	16.21	10.20	1.329	0.695	3.428
FPT	16.00	16.15	18.09	13.02	1.037	−0.495	3.119
RBE	2.759	2.803	7.394	−3.265	2.039	−0.030	2.608
EG	1.867	1.796	13.63	−11.84	3.382	−0.589	5.405
GS	18.20	18.86	26.43	6.532	4.438	−0.417	2.615
IND	26.44	25.20	48.06	16.39	7.143	1.170	3.967
TECH	9.154	9.007	14.33	4.984	1.768	0.627	3.436
FD	4.226	4.471	5.257	2.251	0.681	−0.992	3.188
UR	72.74	78.99	98.15	27.66	16.68	−0.947	3.124
PS	0.269	0.486	1.759	−2.376	0.848	−0.716	2.771

. The range of values for CO₂, FPT, and RBE variables are from 10.20 to 16.21, 13.02 to 18.09, and −3.265 to 7.394, with mean values of 12.60, 16.00, and 2.759, respectively. The means scores for EG, GS, IND, TECH, FD, UR, and PS are (in order) 1.867, 18.20, 26.44, 9.154, 4.226, 72.74, and 0.269, respectively. The SD scores for CO₂, FPT, RBE, EG, GS, IND, TECH, FD, UR, and PS are 1.329, 1.037, 2.039, 3.382, 4.438, 7.143, 1.768, 0.681, 16.68, and 0.848, respectively. The highest maximum score is recorded for UR (98.15) and the lowest minimum is recorded for EG (−11.84). In short, descriptive statistics provide a clear picture of the variables used in the study, which can help in understanding the properties of data and interpretation of the findings of the study. Figure 5 shows a bell-shaped distribution that confirms the data series are normally distributed.

The pairwise correlation estimates between variables are given in

Table 3. Correlation matrix.

	CO2	FPT	RBE	EG	GS	IND	TECH	FD	UR	PS
CO2	1
FPT	0.461	1
RBE	0.387	0.429	1
EG	0.243	0.060	−0.040	1
GS	−0.436	0.325	0.078	−0.289	1
IND	0.287	−0.082	−0.198	0.414	−0.635	1
TECH	0.913	0.540	0.443	0.169	−0.283	0.181	1
FD	0.005	0.514	0.112	−0.068	0.601	−0.259	0.076	1
UR	−0.310	0.087	0.102	−0.344	0.561	−0.606	−0.128	0.012	1
PS	−0.302	0.400	−0.059	−0.212	0.725	−0.520	−0.173	0.478	0.440	1

. It can be seen that CO₂ is positively and strongly attached with FPT coefficient (0.461), with RBE (0.387), with EG (0.243), with IND (0.287), with TECH (0.913), and with FD (0.005). CO₂ is negatively correlated with GS (0.436), with URB (0.310), and with PS (0.302). The strongest correlation is found between TECH and CO₂ (0.913), and the weakest correlation is recorded between IND and GS (−0.635).

5. Empirical Results and Discussion

5.1. Preliminary Results

Before estimating the regression model, preliminary testing is performed to confirm the overall health of the model. Preliminary analysis in our study is based on four steps, namely the CSD, slope homogeneity, unit root, and cointegration tests. The study performed two tests to detect the CSD among selected economies. Our study performed the Pesaran CSD test and the Friedman CSD test for this purpose. See

Table 4. CSD tests.

Variable	CO2	FPT	RBE	EG	GS	IND	TECH	FD	UR	PS
Pesaran’s test	23.36 ***	14.85 ***	13.86 ***	36.82 ***	12.41 ***	13.14 ***	4.150 ***	0.010	6.362 ***	16.49 ***
Friedman’s test	104.6 ***	111.4 ***	102.9 ***	187.3 ***	90.11 ***	120.4 ***	55.53 ***	28.32 ***	56.40 ***	113.7 ***

Note: *** p < 0.01.

for both CSD tests. Both test results reject the null hypothesis of CSD. It shows that datasets for CO₂, FPT, RBE, EG, GS, IND, TECH, FD, UR, and PS are cross-sectionally dependent except FD variable in Pesaran’s test. Slope homogeneity test results are given in

Table 5. Slope heterogeneity test.

	CO2	FPT	RBE	RG	GS	IND	TECH	FD	UR	PS
$\widehat{\mathsf{\Delta }}$	11.22 ***	8.563 ***	9.311 ***	2.419 **	9.857 ***	10.59 ***	9.411 ***	10.83 ***	9.604 ***	7.840 ***
${\widehat{\mathsf{\Delta }}}_{\mathrm{a}\mathrm{d}\mathrm{j}\mathrm{u}\mathrm{s}\mathrm{t}\mathrm{e}\mathrm{d}}$	15.54 ***	11.85 ***	12.89 ***	3.348 ***	13.64 ***	14.66 ***	13.02 ***	15.00 ***	13.29 ***	10.85 ***

Note: *** p < 0.01, ** p < 0.05.

. The null hypothesis of slope homogeneity considers that all slope values are constant and uniform. The slope homogeneity test results reject this null hypothesis and confirm the heterogeneity. In the third step, the study performed a unit root test to confirm the stationarity of variables. For this purpose, the study applied the CIPS unit root test and the outcome is given in

Table 6. CIPS unit root test.

Variable	I(0)	I(1)	Decision
CO2	0.725	−2.720 ***	I(1)
FPT	−1.593	−4.753 ***	I(1)
RBE	−1.874 **		I(0)
EG	−4.558 ***		I(0)
GS	−1.804 *		I(0)
IND	−1.657	−3.923 ***	I(1)
TECH	−1.567	−4.046 ***	I(1)
FD	−1.489	−2.746 ***	I(1)
UR	−1.318	−3.162 ***	I(1)
PS	−2.210 **		I(0)

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

. The CIPS test results show that four variables are level-stationary, whereas the remaining six are first difference stationary. RBE, EG, GS, and PS are level-stationary. CO₂, FPT, IND, TECH, FD, and URB become stationary after taking the first difference. However, none of the variables show stationarity at the second difference.

In the last step, we have confirmed the existence of long-run cointegration between dependent and independent variables. For this, our study used the Kao cointegration test and

Table 7. Kao cointegration test.

	Statistic	p-Value
Modified DF-t	1.553 *	0.060
DF-t	3.095 ***	0.001
Augmented DF-t	2.730 ***	0.003
Unadjusted modified DF-t	2.016 **	0.022
Unadjusted DF-t	3.626 ***	0.000

Note: *** p < 0.01, ** p < 0.05, * p < 0.1.

presents the estimates of the cointegration test. The null hypothesis of this test claims no long-run cointegration exists among the concerned variables. It is confirmed that the null hypothesis is rejected. Thus, this analysis demonstrates that CO₂ emissions and all independent variables are long-term cointegrated.

5.2. Empirical Results

After confirmation of pre-requisite testing, the study proceeds with panel regression analysis. First of all, the study estimates the model with five regression techniques, namely OLS, FE, RE, 2SLS, and system GMM (SGMM).

Table 8. Panel data estimates.

Variables	OLS	FE	RE	2SLS	SGMM
	(1)	(2)	(3)	(4)	(5)
L.CO2					1.024 ***
					(0.006)
FPT	0.182 ***	0.143 ***	0.131 ***	−1.350 ***	−0.036 ***
	(0.033)	(0.038)	(0.039)	(0.301)	(0.004)
RBE	−0.004	−0.024 ***	−0.007	−0.112 ***	−0.005 ***
	(0.012)	(0.007)	(0.007)	(0.026)	(0.001)
EG	0.010	0.007 ***	0.007 **	0.033 ***	0.004 ***
	(0.006)	(0.002)	(0.002)	(0.010)	(0.000)
GS	0.056 ***	0.012	0.008	0.028	0.0003
	(0.009)	(0.008)	(0.008)	(0.017)	(0.0012)
IND	0.016 ***	0.022 ***	0.019 ***	0.045 ***	0.003 ***
	(0.004)	(0.004)	(0.004)	(0.011)	(0.000)
TECH	0.577 ***	0.228 ***	0.333 ***	0.787 ***	0.003
	(0.017)	(0.028)	(0.025)	(0.085)	(0.004)
FD	0.007	0.102 **	0.073 *	0.896 ***	0.023 ***
	(0.046)	(0.042)	(0.044)	(0.193)	(0.006)
UR	0.010 ***	0.0001	0.012 ***	0.002	0.0004 *
	(0.001)	(0.004)	(0.003)	(0.006)	(0.0002)
PS	−0.117 ***	−0.156 ***	−0.113 ***	−0.016	−0.018 ***
	(0.038)	(0.024)	(0.024)	(0.050)	(0.004)
Constant	6.608 ***	7.000 ***	7.411 ***	20.95 ***	0.082
	(0.425)	(0.530)	(0.539)	(2.817)	(0.071)
Observations	529	529	529	529	506
Number of code	23	23	23	23	23

Note: Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

shows the coefficient estimates of all five regression models. FPT estimates are found positive and significant in OLS, FE, and RE models, confirming the positive impact of FPT on CO₂. In contrast, FPT reports a significant decline in CO₂ emission according to 2SLS and SGMM models. These findings display that a 1% upsurge in FPT will increase CO₂ by 0.182% in the OLS model, 0.143% in the FE model, and 0.131% in the RE model, whereas 2SLS and SGMM models results show that a 1% rise in FPT causes 1.350% and 0.036% decline in CO₂ emissions, respectively. RBE estimates are significant and negative in FE, 2SLS, and SGMM models, confirming a negative nexus between RBE and CO₂. These findings report that a 1% rise in RBE will decline CO₂ by 0.024% in the FE model, 0.112% in the 2SLS model, and 0.005% in the SGMM model. The nexus between RBE and CO₂ is statistically insignificant in OLS and RE models.

In the case of control variables, our study reports the increasing impact of EG on CO₂ in all five models; however, the coefficient estimate is statistically insignificant in the OLS model. The coefficient estimates depict that a 1% rise in EG will increase CO2 by 0.007% in the FE and RE models, 0.033% in the 2SLS model, and 0.004% in the SGMM model. GS reports a significant and increasing impact on CO₂ only in the OLS model. Conversely, the impact of GS on CO₂ is statistically insignificant in the remaining four models. The coefficient estimate of GS depicts that a 1% rise in GS will increase CO₂ by 0.056% in the OLS model. IND’s impact on CO₂ was significant and positive in all five models in our study. This shows that industrialization deteriorates environmental quality. The estimates of IND report that a 1% rise in IND activity will raise CO₂ by 0.016% in the OLS model, 0.022% in the FE model, 0.019% in the RE model, 0.045% in the 2SLS model, and 0.003% in the SGMM model. In the case of the technology variable, our results report an increasing impact of TECH on CO₂ in all five models, but the coefficient estimate of the TECH variable is statistically insignificant in the SGMM model. The coefficient estimates of the TECH variable depict that a 1% rise in TECH will increase CO₂ by 0.577% in the OLS model, 0.228% in the FE model, 0.333% in the RE model, and 0.787% in the 2SLS model. FD reports a significant and increasing effect on CO₂ in four models (FE, RE, 2SLS, SGMM). Conversely, the association between FD and CO₂ is found statistically insignificant in the OLS model. The estimates of FD report that a 1% rise in FD will raise CO₂ by 0.102% in the FE model, 0.073% in the RE model, 0.896% in the 2SLS model, and 0.023% in the SGMM model. UR impact on CO₂ is significant and positive in our study’s three models (OLS, RE, SGMM). The estimates of UR report that a 1% rise in urbanization will raise CO₂ by 0.010% in the OLS model, 0.012% in the RE model, and 0.004% in the SGMM model. PS estimates are found significant and negative in the OLS, FE, RE, and SGMM models, confirming a negative impact of PS on CO₂ emissions in highly polluted economies. These findings report that a 1% rise in PS will decline CO₂ by 0.117% in the OLS model, 0.156% in the FE model, 0.113% in the RE model, and 0.018% in the SGMM model. The nexus between PS and CO₂ is found statistically insignificant in the 2SLS model. The coefficient estimates of L.CO₂ are statistically significant and positive, describing how the current level of CO₂ emissions is positively affected by the previous year’s level of CO₂ emissions.

Table 9. Panel quantile regression estimates.

Variables	0.10	0.20	0.30	0.40	0.50	0.60	0.70	0.80	0.90
FPT	−0.014	−0.033	−0.063 *	−0.087 **	−0.114 ***	−0.171 ***	−0.240 ***	−0.283 ***	−0.362 ***
	(0.024)	(0.028)	(0.035)	(0.041)	(0.042)	(0.056)	(0.060)	(0.051)	(0.039)
RBE	0.006	−0.002	−0.001	−0.008	−0.014	−0.023	−0.041 *	−0.060 ***	−0.074 ***
	(0.009)	(0.010)	(0.013)	(0.015)	(0.015)	(0.021)	(0.022)	(0.019)	(0.014)
EG	0.001	0.003	0.004	0.002	0.002	0.011	0.013	0.013	0.007
	(0.004)	(0.005)	(0.006)	(0.007)	(0.008)	(0.010)	(0.011)	(0.009)	(0.007)
GS	0.055 ***	0.064 ***	0.054 ***	0.042 ***	0.034 ***	0.039 **	0.012	0.042 ***	0.066 ***
	(0.007)	(0.008)	(0.010)	(0.012)	(0.012)	(0.016)	(0.017)	(0.015)	(0.011)
IND	0.010 ***	0.012 ***	0.013 ***	0.015 ***	0.011 **	0.016 **	0.012 *	0.015 **	0.019 ***
	(0.003)	(0.003)	(0.004)	(0.005)	(0.005)	(0.007)	(0.007)	(0.006)	(0.005)
TECH	0.759 ***	0.735 ***	0.710 ***	0.687 ***	0.650 ***	0.566 ***	0.529 ***	0.443 ***	0.420 ***
	(0.012)	(0.014)	(0.018)	(0.021)	(0.022)	(0.029)	(0.031)	(0.026)	(0.020)
FD	0.051	0.053	0.006	0.067	0.096	0.078	0.093	0.001	0.050
	(0.035)	(0.039)	(0.049)	(0.057)	(0.059)	(0.078)	(0.083)	(0.070)	(0.055)
UR	0.008 ***	0.008 ***	0.011 ***	0.015 ***	0.017 ***	0.017 ***	0.016 ***	0.014 ***	0.012 ***
	(0.001)	(0.001)	(0.001)	(0.002)	(0.002)	(0.003)	(0.003)	(0.002)	(0.002)
PS	−0.052 *	−0.083 **	−0.073 *	−0.048	−0.023	0.005	−0.085	−0.038	−0.018
	(0.028)	(0.032)	(0.041)	(0.047)	(0.049)	(0.065)	(0.069)	(0.058)	(0.045)
Constant	6.958 ***	6.872 ***	7.064 ***	7.407 ***	7.394 ***	7.478 ***	6.306 ***	6.089 ***	5.832 ***
	(0.311)	(0.365)	(0.456)	(0.526)	(0.546)	(0.721)	(0.766)	(0.653)	(0.508)
Observations	529	529	529	529	529	529	529	529	529

Note: Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

reports the panel quantile regression. Panel quantile regression results report the influence of FPT, RBE, EG, GS, IND, TECH, FD, UR, and PS on CO₂ at the lowest to highest quantiles (0.10th to 0.90th). The FPT variable negatively impacts CO₂ at all quantiles, but the effect is insignificant in the 0.10th and 0.20th quantiles. The RBE suggests a significant negative effect on CO₂ at quantiles 070th to 0.90th. EG positively impacts CO₂ at all quantiles; however, all coefficients are statistically insignificant. Conversely, GS positively impacts CO₂ at lower, medium, and higher quantiles (0.10th to 0.90th), except 0.70th quantile. GS effect on CO₂ is insignificant at 0.70th quantile. IND, TECH, and UR significantly and positively affect CO₂ at all quantiles (0.10th to 0.90th). These findings indicate that industrialization, technology, and urbanization cause detrimental effects on environmental quality. In contrast, the FD effect on CO₂ is found positive at all quantiles (0.10th to 0.90th) but all the estimates are insignificant. However, PS exerts a significant and negative effect on CO₂ only at lower quantiles (0.10th to 0.30th).

5.3. Long and Short-Run Results

Table 10. Long and short-run estimates.

Variable	CS-ARDL	PMG-ARDL
	Coefficient	Coefficient
Long-run
FPT	−0.885 **	−0.545 **
	(0.440)	(0.205)
RBE	−0.089	−0.288 *
	(0.088)	(0.149)
EG	0.028 **	0.478 *
	(0.011)	(0.258)
GS	0.383 *	1.182 *
	(0.210)	(0.656)
IND	0.524	0.005
	(0.465)	(0.033)
TECH	2.975 **	1.899 *
	(1.433)	(0.975)
FD	1.083 *	0.980 *
	(0.541)	(0.579)
UR	0.650	0.472 *
	(1.435)	(0.255)
PS	−0.466	−0.542
	(0.295)	(0.401)
Short-run
D(FPT)	−0.024	−0.053
	(0.112)	(0.039)
D(RBE)	−0.007	−0.021
	(0.022)	(0.021)
D(EG)	0.011 *	0.005 *
	(0.006)	(0.002)
D(GS)	−0.018	−0.011
	(0.037)	(0.008)
D(IND)	0.020	0.005
	(0.029)	(0.007)
D(TECH)	0.362	−0.028
	(0.230)	(0.103)
D(FD)	0.419	0.182 *
	(0.335)	(0.101)
D(UR)	0.603	−0.373
	(0.641)	(0.668)
D(PS)	−0.080	−0.026
	(0.054)	(0.027)
C		−0.056
		(0.148)
ECM(−1)	−0.825 ***	−0.510 ***
	(0.170)	(0.006)
Observations	506	506
Number of countries	23	23

Note: Standard errors in parentheses. *** p < 0.01, ** p < 0.05, * p < 0.1.

presents the long-run (LR) and short-run (SR) estimates of CS-ARDL and PMG-ARDL regression models. A significantly negative association is observed between FPT and CO₂ in the LR in CS-ARDL and PMG-ARDL, with a coefficient estimate of 0.885 and 0.545. It depicts that a 1% upsurge in FPT will lead to 0.885% and 0.545% deterioration in CO₂ in the LR. RBE exhibits an inverse relationship with CO₂ in the LR in both models, but the coefficient estimate is statistically significant in the PMG-ARDL model. The PMG-ARDL model result shows that a 1% rise in RBE will decline CO₂ by 0.288% in the LR. EG is significantly and positively associated with CO₂ in the LR in both models, with coefficients of 0.09902 (CS-ARDL) and 0.0649 (PMG-ARDL). It demonstrates that a 1% increase in EG will encourage CO₂ by 0.028% in the CS-ARDL model and 0.478% in the PMG-ARDL model. GS also reports a significant positive effect on CO₂ in the LR in both models, with coefficients of 0.09902 (CS-ARDL) and 0.0649 (PMG-ARDL). It shows that a 1% rise in GS will boost CO₂ by 0.383% in the CS-ARDL model and 1.182% in the PMG-ARDL model in the LR. On the other hand, IND reports a positive impact on CO₂ in both models in the LR, but the coefficient estimates are insignificant in both models. TECH and FD both variables have significant and positive associations with CO₂ in the LR in both models. These results demonstrate that a 1% increase in TECH will encourage CO₂ by 2.975% in the CS-ARDL model and 1.899% in the PMG-ARDL model, whereas a 1% increase in FD will boost CO₂ by 1.083% in the CS-ARDL model and 0.980% in PMG-ARDL model. UR exhibits a positive relationship with CO₂ in both models in the LR, but the coefficient is statistically insignificant in the CS-ARDL model. The PMG-ARDL model result shows that a 1% rise in UR will raise CO₂ by 0.472% in the LR. On the other hand, PS reports a positive association with CO₂ in both models in the LR, but the coefficient estimates are insignificant in both models.

The SR estimates are presented in the lower panel of Table 10. It is noticed that FPT and RBE both exert a negative impact on CO₂ in the SR in both models, but all the coefficient estimates are insignificant. EG demonstrates a significant and positive influence on CO₂ in both models, with a coefficient of 0.011 (CS-ARDL) and 0.005 (PMG-ARDL). It depicts that a 1% upsurge in EG will boost CO₂ by 0.011% in the CS-ARDL model and 0.005% in the PMG-ARDL model. The FD variable exhibits a positive relationship with CO₂ in the SR in both models, but the coefficient estimate is statistically insignificant in the CS-ARDL model. The PMG-ARDL model result shows that a 1% rise in FD will increase CO₂ by 0.182% in the SR. GS, IND, TECH, UR, and PS coefficient estimates are found insignificant in both models in the SR in highly polluted economies. The error correction model (ECM) estimates are statistically significant and negative in both models, with a coefficient of 0.825 (CS-ARDL) and 0.510 (PMG-ARDL). ECM term reports the speed of adjustment towards equilibrium. The ECM term exhibit that 82% convergence will occur in one year according to the CS-ARDL model and 51% convergence towards equilibrium will occur in one year according to the PMG-ARDL model.

5.4. Results Discussion

The first important outcome suggests the negative influence of FPT, which aligns with our Hypothesis 1. The results suggest that in FPT, the technology and composition effect dominate the scale effect, leading to improved environmental sustainability by controlling carbon emissions. This also implies that the composition effect modifies the basket of finished goods by producing more clean energy items and helping improve the ecosystem. Forest products are normally eco-friendly goods, and their usage can improve environmental sustainability [88]. Trade is the mechanism through which these products can be made available globally. The availability of forest products globally and their increased consumption help reduce carbon emissions and enhance environmental sustainability. Few empirical studies highlight the significance of FPT on carbon footprints. However, many studies believe that forest covers are crucial for promoting environmental quality [89,90]. These studies suggest that forest products are clean and green; thus, the trade could prove beneficial for the ecosystem. However, Khan and Magda [91] contend this finding by confirming the positive influence of forest wood exports on carbon footprints in Asia. On the other hand, Bingpu et al. [92] observed that FPT from China has an insignificant influence on the carbon emissions of Forum on China–Africa Cooperation (FOCAC) members.

The second significant results suggest that bioenergy is crucial in improving environmental quality. Our results supported Hypothesis 2. This is in line with the past studies of Wang [45] and Pathak and Das [49]. Bioenergy is a form of renewable energy that is obtained from organic material and has the potential to positively influence environmental quality by mitigating carbon emissions and reducing dependence on non-renewable or dirty energy sources. Plants and agricultural waste are the biggest sources of bioenergy, which has the ability to absorb carbon emissions during their growth. Thus, in comparison to traditional energy sources, increasing the use of bioenergy results in a lower net upsurge in CO₂ concentration in the atmosphere. In order to contribute to environmental sustainability, a carbon-neutral cycle produced by bioenergy sources is crucial. Moreover, waste materials are also an important source of bioenergy, helping reduce landfill use and associated methane emissions. Therefore, increasing the share of renewable energy sources in the total energy mix can help achieve a cleaner and greener environment. These empirical inferences are also supported by Gao and Zhang [51].

6. Conclusions and Implications

6.1. Conclusions

Environmental sustainability is the primary objective of policymakers all around the globe. As non-renewable or conventional energy sources fuel the rising economic activities, the biggest source of carbon emissions is believed to be the biggest hurdle in achieving environmental sustainability. The most viable option to deal with this situation is to increase the use of renewable energy sources, particularly bioenergy, a carbon-neutral source of energy. In addition to bioenergy, trading activities in clean and green products can also prove vital in enhancing environmental performance. The literature is growing with regard to the impact of rural bioenergy and trade in environmental goods on environmental sustainability. However, empirical evidence on the impact of FPT and rural bioenergy on environmental sustainability has not been found in the contemporary literature. Therefore, the primary research question this analysis wants to address is whether FPT and rural bioenergy can help environmental sustainability. The analysis intends to answer this question by examining the impact of FPT and rural bioenergy on environmental sustainability. Empirical estimates of the model are obtained by applying several estimation techniques, such as OLS, FE, RE, 2SLS, and GMM.

The findings of these regression techniques are as per our expectations. The estimated coefficients of FPT are positively connected to CO₂ emissions in the OLS, FE, and RE, and negatively connected to CO₂ emissions in 2SLS and GMM models. However, the estimates of bioenergy are negatively linked to CO₂ emissions in FE, 2SLS, and GMM models, suggesting that bioenergy helps improve environmental sustainability. The estimates of control variables of economic growth, industrialization, technological development, urbanization, and financial development are positively significant, confirming that these factors increase carbon footprints and thus hurt environmental sustainability. In contrast, political stability negatively impacts carbon emissions and thus promotes environmental sustainability. The CS-ARDL and PMG-ARDL results show that FPT and bioenergy have a favorable impact on environmental sustainability by reducing CO₂ in the long run.

6.2. Policy Implications

Relying on the primary findings, we provide some important policy guidelines, which are as follows:

• Our findings suggest that the FPT significantly reduced carbon footprints and improved environmental quality. It is suggested that policymakers take concrete steps to enhance the trade of forest products. In this regard, policymakers should follow Forest Stewardship Council guidelines when carrying out FPT. These certifications play an important role in promoting the forest products that are developed via sustainably managed forests, taking into account the protection of biodiversity and maintaining ecosystem services. In order to promote FPT, policymakers should provide financial incentives to those traders who follow sustainable trading practices and firms involved in forest management. Further, policymakers should enforce stringent policies crucial in reducing illicit logging and the sale of forest products. Thus, strengthening the legislative structure is crucial for promoting forest sustainability, which can reduce deforestation in the case of increased FPT.
• Bioenergy is a crucial factor in promoting the sustainability of the ecosystem; therefore, in order to reduce emissions, we suggest that policymakers design policies that facilitate the transition towards renewable energy sources. However, a cautious approach should be adopted during the transition process because replacing fossil fuels with renewable energy sources is a steady process that may slow down the pace of economic activities. Moreover, the transition process should be facilitated, and its speed can be increased by facilitating public-private partnerships in the renewable energy sector. Further, specifically designed laws and policies that address delays and lags in the bioenergy industry can remove administrative hurdles in the way of developing bioenergy projects. This can enhance the pace of the process toward renewable energy transition and enhance the proportion of renewable energy sources, particularly bioenergy, in total energy sources.

6.3. Limitations and New Directions

This study has made several valuable contributions, but a few weaknesses are worth mentioning. First, the study relies on linear analysis; however, the macro variables mostly behave asymmetrically. Particularly, the variables in business cycles move non-linearly. The variables FPT and bioenergy are vulnerable to external shocks and incorporating them in a linear model and estimating them with linear techniques can lead to erroneous estimates. Therefore, we suggest applying nonlinear techniques to assess the impact of FPT and bioenergy on environmental sustainability. Second, the study has selected CO₂ emissions to represent environmental quality; however, CO₂ emissions do not fully represent environmental sustainability. Future studies should consider ecological footprint as a proxy of environmental sustainability, a comprehensive measure of environmental quality. Third, the comparative analysis between advanced and emerging economies can add more value to future analysis.