Reinvigorating the role of clean energy transition for achieving a low-carbon economy: evidence from Bangladesh

Abstract

Achieving carbon-neutrality has become a global agenda following the ratification of the Paris Agreement. For the developing countries, in particular, attaining a low-carbon economy is particularly important since these economies are predominantly fossil-fuel dependent, to which Bangladesh is no exception. Therefore, this study specifically aimed at evaluating the environmental impacts associated with energy consumption and other key macroeconomic variables in the context of Bangladesh over the 1975–2016 period. As opposed to the conventional practice of using carbon dioxide emissions to proxy environmental quality, this study makes a novel attempt to use the carbon footprints to measure environmental welfare in Bangldesh. The outcomes from this study are expected to facilitate the carbon-neutrality objective of Bangladesh and, therefore, enable the nation to comply with its commitments concerning the attainment of the targets enlisted under the Paris Agreement and the United Nations Sustainable Development Goals declarations. The econometric analysis involved the application of methods that are suitable for handling the structural break issues in the data. The overall findings from empirical exercises reveal that aggregate energy consumption, fossil fuel consumption, and natural gas consumption boost the carbon footprint figures of Bangladesh. In contrast, nonfossil fuel consumption and hydroelectricity consumption are witnessed to abate the carbon footprint levels. Besides, economic growth and international trade are also evidenced to further increase the carbon footprints. Hence, these findings suggest that a clean energy transition within the Bangladesh economy can be the panacea to the nation's persitently aggravating environmental hardships. Furthermore, the causality analysis confirmed the presence of unidirectional causalities stemming from total energy consumption, fossil fuel consumption, natural gas consumption, hydroelectricity consumption, economic growth, and international trade to the carbon footprints. On the other hand, nonfossil fuel consumption is found to be bidirectionally associated with carbon footprints. In line with these aforementioned findings, several key policy suggestions are put forward regarding the facilitation of the carbon-neutrality agenda in Bangladesh.

Introduction

Achieving carbon-neutrality has become a global agenda following the ratification of the Paris Agreement (Qin et al. 2021; Ma et al. 2021; Liu et al. 2021). Consequently, the contemporary studies in the environmental economics narrative have documented evidence regarding the environmental effects associated with changes in various macroeconomic aggregates (Al-mulali et al. 2016; Rauf et al. 2018; Iheonu et al. 2021). It is believed that attaining a low-carbon economy is particularly important for developing nations since these countries are predominantly fossil fuel dependent. Similarly, Bangladesh is a rapidly developing country from the Indian subcontinent which is yet to phase out its traditional reliance on fossil fuels for power-generation purposes (Murshed 2021a,b). Bangladesh, in respect of its economic performances, has fared well in transitioning from being characterized as a ‘bottomless basket’ to be considered as an ‘economic role model’ for similar world economies (Barai 2020). Such a remarkable growth explosion can be credited to the nation’s notable degrees of socioeconomic development over the last couple of decades. As per the World Bank (2020) estimates, the Bangladesh economy has been growing at a rate of over 5%, on average per year, during the last decade. As a result, Bangladesh has recently overtaken India to emerge as the fastest growing South Asian economy. Besides, the nation has also recently attained the eligibility criteria required to graduate from the Least Developed Countries (LDC) group to the Lower-Middle-Income Countries (LMIC) group by 2024. Moreover, the Goldman Sachs Investment Bank has classified Bangladesh as one of the Next Eleven (N11) nations that are poised to achieve substantial degrees of economic growth in the 21^st century. Among the N11 nations, Bangladesh accounted for the second-highest real GDP growth rate of 6.4% in 2015 (Kuepper 2019, May 06). Furthermore, the nation is presumed to become the epicenter of global investments in the future whereby the size of the Bangladesh economy is likely to be further expanded over the next decade.

However, despite flourishing economically, the environmental traditional performances of Bangladesh have been unsatisfactory to a large extent (Murshed, 2020a, b; Murshed and Dao 2020). In 2019, Bangladesh’s Carbon dioxide (CO₂) emissions per capita were around 0.6 metric tonnes which have risen by around 3-fold compared to the CO₂ emission level in 2000 (Friedlingstein et al. 2020). It is to be noted that almost two-thirds of Bangladesh’s total CO₂ emissions are derived from gaseous fossil fuel consumption (World Bank 2020) which portrays the nation’s predominant dependency on fossil fuels (Islam et al. 2021; Murshed et al. 2021a). On the other hand, from 1990 onwards, the total greenhouse gas emissions in Bangladesh have increased by more than 45% (World Bank 2020). On the other hand, Bangladesh is also currently enlisted as one of the global nations with the worst possible air quality; consequently, the capital city Dhaka is the second-most polluted city across the world (IQ Air 2019). Although Bangladesh emits relatively less amount of CO₂ per capita compared to many of the developed world economies (Ahiduzzaman and Islam 2011), its CO₂ emission figures do not entirely reflect the multidimensionality of the environmental hardships it has endured in the past. Environmental pollution in Bangladesh is not only limited to greenhouse gas emissions-induced air pollution. For instance, the water quality in Bangladesh has persistently deteriorated primarily due to the disposal of industrial effluents, domestic waste, and agricultural runoffs (Hassan et al. 2019). Similar anthropogenic factors have also deteriorated the land quality in Bangladesh (Islam et al. 2019). Besides, deforestation is another of the major forms of environmental degradation in Bangladesh since the nation is extremely vulnerable to natural calamities like floods and landslides (Islam et al. 2019; Murshed et al. 2021b). Hence, addressing the factors responsible for the aggravation of the environmental quality in Bangladesh has become a key policy agenda of the government.

Achieving a low-carbon economy is of paramount relevance for Bangladesh particularly due to its global commitments to adopt relevant policies for abating CO₂ and other greenhouse gas emissions. Bangladesh is among the 196 global economies that have ratified the Paris Agreement at the United Nations Climate Change Conference (COP21) in December 2015 (Li et al. 2021; Rehman et al. 2021). Consequently, the nation has a responsibility of gradually mitigating its emissions to collectively limit global warming this century to below 1.5–2° C above the pre-industrial levels. At the same time, Bangladesh has also committed to achieving the targets enlisted under the United Nations 2030 Sustainable Development Goals (SDG) agenda which calls for simultaneously achieving social, economic, and environmental sustainability across the world. Hence, focusing on the environmental sustainability goal, it is critically important for Bangladesh to pursue its carbon-neutrality strategies especially through the greening of its production and consumption processes. Therefore, it is clear that achieving carbon-neutrality is of paramount significance for Bangladesh, especially for the nation to comply with the associated international commitments. Besides, it is also of local interest for Bangladesh since the nation is one of the global nations that are highly susceptible to climate change atrocities (Hasnat et al. 2018). Besides, reducing CO₂ emissions can also be considered as a national agenda since the emission-induced climate change consequences have been acknowledged to negatively affect the domestic sectors as well, especially agriculture (Shakoor et al. 2021; Dagar et al. 2021). Since the agriculture sector makes a decent contribution to Banglades's total value added, achieving carbon-neutrality can also be effective in facilitating the nation's food security agenda. Most importantly, transforming from a pollution-intesive to a low-carbon economy would ultimately enable Bangladesh to simultaneously achieve both economic and environmental welfare in the years to come. 

Among the major factors responsible for environmental degradation in Bangladesh, the nation’s monotonic fossil fuel dependency, in the past and also at present, is said to have adversely impacted the environmental quality (Murshed et al. 2021c). The nation has predominantly utilized both local and imported fossil fuels to bridge its persistently increasing energy demand; consequently, a major portion of Bangladesh’s total final energy consumption level is currently generated from the combustion of fossil fuels (World Bank 2020; Murshed and Alam 2021). During the post-2000 period, fossil fuels on average accounted for almost two-thirds of Bangladesh’s aggregate primary energy consumption figures (World Bank 2020). On the other hand, the dismal state of fossil fuel dependency of the nation can be further understood from the low renewable electricity output shares of Bangladesh. Consequently, such monotonic fuel dependency has resulted in persistent aggravation of the environmental indicators. Under such circumstances, it is has been prescribed that the nation should gradually phase out its fossil fuel dependency through the addition of renewable energy resources into the national energy mix (Murshed and Tanha 2020). Hence, keeping environmental sustainability into cognizance, a Clean Energy Transition (CET)¹ is extremely important for Bangladesh. However, the nation is yet to succeed in enhancing its shares of renewables in both the total primary energy consumption and electricity output levels (World Bank 2020). Rather, the fossil fuel dependency of Bangladesh has amplified with time (Murshed et al. 2021d). Moreover, the government of Bangladesh, to bridge the national energy crisis, is likely to tap the domestic coal supplies to generate electricity (Islam and Khan 2017; IEA CCC 2019). Such energy policy decisions can be expected to deteriorate the environment in Bangladesh further Howladar (2013). Against this backdrop, this paper aims to empirically identify the macroeconomic factors which can facilitate Bangladesh’s carbon-neutrality objective and enable the nation to comply with its Paris Agreement and SDG commitments.

This current study contributes to the existing literature in several aspects. Firstly, keeping the carbon-neutrality agenda into consideration, the Carbon Footprints (CFP) are used to measure the quality of the environment in Bangladesh. Compared to CO₂ emissions, the CFP are more comprehensive indicators of CO₂-induced environmental degradation since these footprints are measured in terms of the quantity of productive land area needed to absorb the CO2 emissions stemming from the combustion of fossil fuels, in particular Wackernagel et al. (2004). The preceding studies in the context of Bangladesh have predominantly used the CO₂ emission figures to proxy for environmental quality (Sarkar et al. 2018; Baul et al., 2018). However, very few studies have attempted to use the CFP, and, to some extent, also the overall Ecological Footprints (EFP)², in this regard. The relevance of using the CFP, as opposed to using CO₂ emissions to measure environmental quality, is derived from the understanding that the CFP tend to provide a more comprehensive representation of the changes in the overall quality of the environment.³ Secondly, to assess the possible heterogeneous impacts of different energy resource consumption levels on the environment, this study separately evaluates the impacts of fossil and nonfossil fuels, both at aggregate and disaggregated levels, on the CFP figures of Bangladesh. To the best of the authors’ knowledge, this is the only study that specifically scrutinizes the relationships between CFP and energy consumption for the case of Bangladesh. Besides, for robustness check, the impacts of natural gas and hydroelectricity consumptions, to respectively proxy for fossil and nonfossil fuel uses, on the CFP figures are also explored. As opposed to the majority of the existing studies that have separately assessed the fossil fuel consumption-environmental quality and the nonfossil fuel consumption-environmental quality nexuses in Bangladesh’s context, this study provides scope for comparison between the environmental impacts associated with the uses of these alternative energy resources. Lastly, this study adds to the limited literature that has overlooked the issues of Structural Breaks (SB) in ascertaining the energy consumption-environmental quality nexus for Bangladesh. Since Bangladesh has experienced several episodes of macroeconomic, political and natural calamity shocks, it is pertinent to control for the possible SB in the data. Therefore, as the preceding studies have not accounted for these concerns within their estimation processes, the findings document in the literature can be presumed to be biased. Consequently, this current study addresses this literature gap by employing econometric methods that are robust to handling multiple SB breaks in the data.

The remainder of the paper is structured as follows. “The trends in the CFP and EFP of Bangladesh” section briefly discusses trends in the CFP and EFP figures of Bangladesh. Besides a snapshot of the renewable energy consumption scenario of Bangladesh is also highlighted in this section. The “Literature review” section presents a review of the relevant literature. The empirical models and the attributes of the data set used in this study are specified in the “Empirical models and data” section. The econometric methodology is explained in the “Methodology” section. In the “Results and Discussions” section, the empirical findings are reported and discussed. Finally, “Conclusion and policy implications” section concludes and suggests relevant policy interventions.

The trends in the CFP and EFP of Bangladesh

Following its independence, the size of the Bangladesh economy has substantially enlarged whereby higher economic growth has simultaneously degraded the environment. The persistent deterioration of the environmental quality in Bangladesh can be understood from the trends in the nation’s EFP figures shown in Fig. 1. It can be seen that Bangladesh has been ecologically deficit throughout the 1975–2016 period. The EFP per capita figures have increased by more than 1.5 times while the biocapacity of Bangladesh in meeting the diversified human demands for natural resources has declined. Consequently, the ecological deficits have persistently gone up. Over the 1975–2016 period, the per capita ecological deficits have increased by almost 4.5 times. As far as the trends in the carbon and noncarbon footprints are concerned, Fig. 2 shows that both these components of the total EFP have risen over the years. However, compared to the growth of the per capita noncarbon footprints, the per capita CFP figures of Bangladesh have substantially amplified over the aforementioned period. This is the major motivation behind this study.

Fig. 1.

The trends in the Ecological footprints of Bangladesh (1975–2016)

Fig. 2

The trends in the carbon and noncarbon ecological footprints of Bangladesh (1975–2016)

Notes: The total ecological footprints, biocapacity, and ecological deficit are based on the footprints of consumption

Source: GFN (2019)

Notes: The carbon ecological footprints are presented along the secondary axis

Source: GFN (2019)

The trends in the energy consumption levels in Bangladesh

The impacts of energy use on the environmental indicators are assumed to be conditional on the type of energy resource employed. Traditionally, fossil fuels have been a predominant source of energy employed for producing the national output of Bangladesh. The combustion of fossil fuels has been criticized for stimulating unparalleled degrees of environmental pollutants into the atmosphere. As a result, undergoing CET has become a vital mechanism for ensuring environmental sustainability in Bangladesh (IRENA 2017). However, the nation is yet to significantly increase its overall renewable energy shares in the total final energy consumption and electricity output figures. Figure 3 illustrates the energy consumption trends in Bangladesh over the 1990–2015 period. It can be seen that renewable energy accounted for almost 72% of the total final energy consumption figures of Bangladesh in 1990. However, ever since, this share has relentlessly declined. Between 1990 and 2015, renewable energy shares have reduced by almost 37 percentage points. On the other hand, the state of the renewable electricity outputs of Bangladesh has been even worse. In 2015, merely 1.23% of the total electricity outputs of Bangladesh were generated using renewable resources, hydroelectricity in particular. Besides, between 1990 and 2015, renewable electricity outputs have declined by a staggering 10.2% percentage points. Currently, Bangladesh generated the bulk of its electricity outputs using the indigenous natural gas reserves. However, following the acute shortages in the natural gas supply, the government is planning to set up coal-fired power plants to meet the electricity demand. However, diversifying the energy mix in this manner would boost the fossil fuel dependency of the nation whereby CET in Bangladesh can be expected to be further delayed. Hence, the nation must rethink its decision of introducing coal into the national energy basket. Simultaneously, renewable energy resources are ideally to be augmented into the energy systems which, in turn, can be anticipated to reinstate environmental welfare in Bangladesh.

Fig. 3.

The trends in renewable and nonrenewable energy and electricity consumptions in Bangladesh

Source: World Bank ( 2020 )

Literature review

The preceding studies focusing on the impacts of energy consumption on the environment have predominantly evaluated the effects of aggregate energy consumption on CO₂ emissions (Ben Jebli et al. 2015; Farhani and Ozturk 2015; Alam et al. 2016). In a study on Turkey, Ozturk and Acaravci (2010) scrutinized the energy consumption-CO2 emissions nexus using annual data from 1968 to 1995. The authors concluded that a rise in the national income level is associated with a decline in the Turkish CO₂ emissions figures. Similarly, Wang et al. (2016) found a long-run association between energy use, CO₂ emissions, and economic growth in China between 1990 and 2012. Besides, the authors also asserted that energy consumption causally influences CO₂ emissions without the feedback association. Nain et al. (2017) referred to electricity consumption as a short-run influencer of higher CO₂ emissions in India. However, sectoral electricity consumption was found to exert contrasting causal associations with India’s CO₂ emission figures. In a study by Mirzaei and Bekri (2017) on Iran, the authors predicted a positive correlation between energy consumption and CO₂ emissions. In line with their findings, the authors concluded that energy consumption and emissions of CO₂ in Iran are likely to increase by 4.3% and 5%, respectively, between 2010 and 2050. Among the cross-sectional studies, Wang et al. (2018) found a long-run association between energy consumption and CO₂ emissions for 170 low, lower-middle, upper-middle, and high-income economies. Moreover, the authors also found evidence of heterogeneous causal nexuses between energy consumption and CO₂ emissions across the different income groups. Identical results were also put forward by Chen et al. (2016) in the context of 188 developing and developed economies. Similarly, Muhammad and Khan (2021) used data of 170 global economies and found statistical evidence of higher energy consumption resulting in environmental degradation by catalyzing the CO₂ emission levels. In another recent study on seven emerging economies, Tong et al. (2020) found evidence of bidirectional short-run causation between total primary energy consumption and CO₂ emissions for all seven nations apart from Indonesia. In the context of Indonesia, a reverse causality running from CO₂ emissions to total primary energy consumption was found in the short run.

The energy consumption-CO₂ emission nexus was also evaluated using disaggregated energy consumption data (Murshed, 2021c). The findings from these studies have collectively referred to the relationship between energy consumption and CO₂ emissions being conditional on the type of energy resource consumed (Dong et al. 2017a; Murshed, 2020a, b). Saboori et al. (2017) explored the CO₂ emission responses to oil consumption in three oil-importing Asian economies (Japan, South Korea, and China) over the 1980–2013 period. The empirical results indicated that positive shocks to oil consumption levels reduced CO₂ emissions in China and Japan. However, in the case of South Korea, higher consumption of oil initially resulted in lower CO₂ emissions while in the long run the relationship was reversed. The results from the causality analysis revealed unidirectional causation extending from oil consumption to CO₂ emissions only in the context of China and South Korea. Recently, Khan et al. (2020) investigated the impacts of nonrenewable fossil fuel consumption on Pakistan’s CO₂ emissions between 1965 and 2015. The results implicated that consumption of coal, natural gas, and oil accounted for higher emissions of CO₂ in the country both in the short- and the long run. In contrast, Dong et al. (2017b) opined in favor of natural gas consumption reducing CO₂ emissions across 30 Chinese provinces during the 1995–2014 period.

On the other hand, several studies have also addressed the impacts of renewable nonfossil fuels on CO₂ emissions. Bilgili et al. (2016) found the consumption of renewable energy resources to lower down CO₂ emissions in the context of 17 Organization for Economic Cooperation and Development (OECD) member countries. Similar findings were documented by Al-Mulali et al. (2016) in the context of Western European, East Asian and the Pacific, South Asian, and North American nations. In other relevant country-specific studies, Zhang et al. (2017) found renewable energy consumption and CO₂ emissions to be negatively correlated for Pakistan while Sinha and Shahbaz (2018) referred to renewable electricity output accounting for lower CO₂ emissions in India. Similarly, highlighting the favorable role of nonfossil fuel consumption on the betterment of environmental quality in China, Zhao et al. (2017) predicted that replacing the consumption of coal with wind power is likely to significantly reduce the nation’s CO₂ emission figures in the future. Likewise, the consumption of solar power has also been identified as a means of curbing CO₂ emissions (Lodi et al. 2018). In contrast, Nathaniel et al. (2021) recently documented evidence of renewable energy consumption not being able to explain the variations in the CO2 emission levels in the G7 countries; however, the authors did conclude that nuclear energy use is a credible means of abating emissions in these countries. Hence, these aforementioned studies have mostly identified fossil fuel consumption to be detrimental to environmental quality. In contrast, nonfossil fuels have been associated with improvements in the quality of the environment.

Although the initial studies on the energy consumption-environmental quality nexus have predominantly used the CO₂ emission levels to proxy for environmental quality, contemporary studies have preferred to use the EFP to quantify the state of the environment (Nathaniel 2021; Solarin et al. 2021). In a study on Qatar, Charfeddine (2017) found evidence of electricity consumption leading to higher EFP. It is to be noted that Qatar is an exporter of oil and its electricity outputs are mostly generated from combusting the domestic oil supplies which, in turn, could have influenced the findings reported in that study. In another study on 44 African nations, Nathaniel and Adeleye (2021) remarked that energy consumption is detrimental to achieving environmental sustainability across Africa since higher energy consumption was evidenced to boost the EFP of these nations. Similarly, in the context of 11 newly industrialized economies, Destek and Sarkodie (2019) found that energy consumption accentuated the EFP across these countries between 1977 and 2013. However, the country-specific analysis statistically certified the positive energy consumption-EFP nexus for China, India, Mexico, Singapore and Thailand but not for Brazil, Malaysia, Philippines, South Africa, South Korea, and Turkey. The positive energy consumption-EFP nexus was also presented in the study by Baloch et al. (2019) in the context of 59 countries under China’s Belt and Road Initiative (BRI) which included Bangladesh as well.

In another relevant study featuring 16 European Union member nations over the 1997–2014 period, Alola et al. (2019) showed that both renewable and nonrenewable energy consumption had positive long-run associations with the EFP. However, the marginal impacts of renewable energy consumption, in this regard, were found to be significantly lower than that of nonrenewable energy consumption. Hence, these results suggested that renewable energy is a relatively environmentally friendly fuel. Destek and Sinha (2020) also reported the dissimilar effects of renewable and nonrenewable energy use on the EFP in the context of 24 OECD nations. Using annual data from 1980 to 2014, the authors claimed that greater consumption of renewable energy reduced the EFP while nonrenewable energy consumption was found to increase the EFP. The role of energy consumption on the EFP of Malaysia was assessed by Ahmed et al. (2019). The empirical results indicated that only in the long-run energy consumption increased the EFP in Malaysia. Identical results were put forward by Ahmed and Wang (2019) in the context of India. However, Nathaniel et al. (2019) found energy use in South Africa to increase and decrease the nation’s EFP in the short- and long run, respectively. Thus, the authors highlighted that energy consumption, especially in the form of renewable energy resources, in the long run, would lead to better environmental quality in South Africa.

Among the more recently published studies on the energy consumption-EFP nexus, Ulucak and Khan (2020) used annual data from 1992 to 2016 in the context of the BRICS⁴ nations and found renewable energy consumption can be the panacea to the environmental degradation concerns of these emerging economies by reducing their respective EFP figures. Besides, disaggregating overall energy consumption into oil and gas consumption, Majeed et al. (2021) claimed that positive shocks to the consumption levels of these nonrenewable fossil fuels lead to a rise in the EFP levels of Pakistan. Moreover, the authors also found evidence of a decline in oil and gas consumption being efficient in reducing Pakistan’s EFP figures. In another recent study on Turkey, Kirikkaleli et al. (2021) stated that boosting energy consumption could facilitate Turkey’s economic growth objectives but it would hamper the nation’s environmental quality by boosting the EFP figures both in the short- and long run. Although the majority of the existing studies have examined the effects of energy use on the total EFP figures, a limited number of studies have explored the effects on CFP. Ulucak and Danish (2020), in the context of the OECD nations, documented evidence of renewable energy consumption helping to reduce the CFP figures of these countries. Furthermore, the authors also asserted that the transition from nonrenewable to renewable energy within these nations can significantly help to improve the environmental quality since fossil fuel consumption was evidenced to boost the CFP figures.

The literature on the energy consumption-environmental quality nexus for Bangladesh

The previous studies on the impacts of energy consumption on environmental quality in Bangladesh have also been predominantly centered on impacts on CO₂ emissions in particular. In a study by Alam et al. (2012), the authors found a long-run association between aggregate energy consumption and CO₂ emissions in Bangladesh over the 1972–2006 period. Besides, bidirectional causations between these variables, both in the short- and the long- run, were ascertained. In a similar study for Bangladesh between 1972 and 2011, Rahman and Kashem (2017) found a cointegrating association between the growth of per capita energy use and per capita CO₂ emissions. However, the results indicated a unidirectional causality stemming from energy consumption to CO₂ emissions. Moreover, the regression findings statistically certified the positive relationship between these variables both in the short- and long- run. Similar findings were ascertained in the study by Oh and Bhuyan (2018). In contrast, Khan (2013) could not establish a statistically significant causal linkage between energy consumption and CO₂ emissions using annual data from 1965 to 2007 in the context of Bangladesh. Recently, Ahmed et al. (2020) probed into the energy consumption-CO₂ emission nexus between 1972 and 2010. However, no causal associations between these variables could be statistically validated.

The effects of energy use on CO2 emissions were also analyzed through the lens of the Environmental Kuznets Curve (EKC) hypothesis (Apergis and Ozturk 2015; Koondhar et al. 2021; Alola and Ozturk 2021). The relevant studies in this regard have mostly claimed that the nonlinear inverted U-shaped relationship between economic growth and CO2 emissions is conditional on the level of energy utilized within the economy (Al-mulali and Ozturk 2016; Le and Ozturk 2020). In a relevant study for Bangladesh, over the 1971–2010 period, Islam et al. (2013) employed the EKC model and found total energy consumption to adversely impact the nation’s CO₂ emissions both in the short- and the long run. The authors also remarked that energy consumption had a critical role to play to bend the EKC for Bangladesh. Similarly, Nasreen et al. (2017) also explored the impacts of energy consumption on CO₂ emissions under the EKC hypothesis-testing framework. The study considered annual data from 1980 to 2012 for five South Asian economies including Bangladesh. In line with the findings, the authors concluded that in the long run energy consumption contributed to higher CO₂ emissions across all five South Asian countries. In another study involving these five South Asian economies, Akhmat et al. (2014) reported unidirectional long-run causality running from electricity consumption to CO₂ emissions. Shahbaz et al. (2014) also found statistical evidence regarding higher per capita electricity consumption, between 1975 and 2010, increasing Bangladesh’s CO₂ emissions in the long run. However, in the short run electricity consumption was claimed to reduce CO₂ emissions in Bangladesh.

Apart from the country-specific examinations, the energy consumption-CO₂ emissions nexus concerning Bangladesh was also explored using panel datasets. Among these, Padhan et al. (2019) found energy consumption to catalyze CO₂ emissions across the panel of the N11 economies. Similarly, Nasreen et al. (2017) concluded that energy consumption and financial stability degrade environmental quality by boosting CO₂ emissions in the South Asia countries. In a similar study using disaggregated energy consumption data for the N11 nations, Sinha et al. (2017) concluded that renewable energy consumption reduced CO₂ emissions while biomass and nonrenewable energy consumption increased CO₂ emissions. On the other hand, using panel data of five South Asian countries (Bangladesh, India, Pakistan, Sri Lanka, and Nepal), Siddique et al. (2016) found a positive correlation between total energy consumption and CO₂ emissions. Besides, in the short run, a unidirectional causality running from CO₂ emissions to energy consumption was also evidenced. A similar correlation between energy consumption and CO₂ emissions concerning these five South Asian economies was also asserted in the study by Zakaria and Bibi (2019).

Although several studies investigated the impacts of energy use on CO₂ emissions, no country-specific study has been conducted to assess the energy consumption-CFP nexus in the Bangladesh context. However, panel estimations involving Bangladesh in this regard have been carried out using the EFP to proxy environmental quality. In a study on 13 Asian economies including Bangladesh, Lu (2020) highlighted that since the energy mix of the Asian countries is predominantly fossil fuel-intensive, a rise in the overall energy consumption level adds to the EFP figures of these nations. Baloch et al. (2019) found evidence regarding higher energy consumption, within 59 countries under the BRI of China, attributing to the rise in the EFP. In another relevant study on the N11 economies, Sinha et al. (2019) concluded that renewable energy consumption contributed to lower down the EFP while fossil fuel consumption was found to boost the EFP.

Therefore, it is apparent from the aforementioned empirical evidences on Bangladesh that the use of the CFP to assess the energy consumption-environmental quality nexus is yet to be extensively explored. Moreover, most of the existing studies have predominantly assessed the impacts of aggregate energy consumption on Bangladesh’s environmental quality. However, the possible fuel-specific heterogeneous impacts of energy consumption on the environmental attributes need to be evaluated for policy implications. Hence, this paper addresses these gaps in the literature by exploring the impacts of both fossil and nonfossil fuel consumption on the CFP of Bangladesh.

Empirical models and data

Following Nathaniel and Khan (2020), the environmental impacts of energy consumption in Bangladesh are modeled using a double-log linear function in which the CFP figures of Bangladesh are expressed as a function of energy consumption and other key control variables including economic growth, urbanization, and international trade. The choice of the control variables is in line with the theoretical frameworks highlighted in the relevant literature. The baseline model can be specified as:

$$ {\mathrm{lnCFP}}_{\mathrm{t}}={\partial}_0+{\partial}_1{\mathrm{lnEN}}_{\mathrm{t}}+{\partial}_2{\mathrm{lnRGDP}}_{\mathrm{t}}+{\partial}_3{\mathrm{lnURB}}_{\mathrm{t}}+{\partial}_4{\mathrm{lnOPEN}}_{\mathrm{t}}+{\upvarepsilon}_{\mathrm{t}} $$

(1)

where the subscript t denotes the time period (year), ε_t is the residual error-term, ∂₀ is the intercept and ∂_i(i = 1, …, 4) are the elasticity parameters to be predicted. The dependent variable CFP refers to the per capita CFP of Bangladesh. Higher levels of per capita CFP indicate deterioration of the environmental quality and vice-versa (Rees 1992; Wackernagel et al. 2004). The variable EN refers to the total energy consumption per capita levels. Depending on the pattern of the impacts of energy use on the environment, ∂₁ can be expected to either depict a negative sign (Destek and Sarkodie 2019) or a positive sign (Al-Mulali et al. 2015).

The empirical model is controlled for the level of economic growth within the Bangladesh economy since the quality of the environment is believed to be influenced by the economic activities which take place along the growth stream (Liang and Yang 2019). The variable RGDP refers to the real GDP level of Bangladesh. Several studies have used real national income levels to proxy for the level of economic growth (Nathaniel and Khan 2020; Ahmed et al. 2021). Considering the possible heterogeneity of the environmental impacts of economic growth, the sign of the elasticity parameter ∂₂ could also be either positive or negative (Sharif et al. 2020). The variable URB abbreviates for the rate of urbanization within the Bangladesh economy. The inclusion of the urbanization rate into the model is justified from the understanding that the unplanned urbanization woes of Bangladesh have accounted for unprecedented environmental hardships for the city-dwellers in particular (Subrina and Chowdhury 2018). Thus, the elasticity parameter ∂₃ can be expected to bear a positive sign. The urbanization-EFP nexus was explored in several preceding studies (Nathaniel et al. 2019, 2020). Finally, the model is also controlled for the impacts of international trade flows on the CFP. Following Charfeddine 2017 and Destek and Sinha (2020), international trade is proxied by the trade openness index (OPEN). Since international trade in developing countries, in particular, is expected to induce specialization in pollution-intensive production processes, the sign of the elasticity parameter ∂₄can be expected to be positive (Ali et al. 2020).

The CFP, and also the overall EFP, are often claimed to be influenced by the type of fuel consumed (Holden and Høyer 2005). Hence, to account for the possible fuel-specific impacts of energy use on the environment, model (1) is reestimated using disaggregated levels of fossil and nonfossil fuel consumption. The modified models can be expressed as:

$$ {\mathrm{lnCFP}}_{\mathrm{t}}={\updelta}_0+{\updelta}_1{\mathrm{lnFOS}}_{\mathrm{t}}+{\updelta}_2{\mathrm{lnRGDP}}_{\mathrm{t}}+{\updelta}_3{\mathrm{lnURB}}_{\mathrm{t}}+{\updelta}_4{\mathrm{lnOPEN}}_{\mathrm{t}}+{\upvartheta}_{\mathrm{t}} $$

(2)

$$ {\mathrm{lnCFP}}_{\mathrm{t}}={\uprho}_0+{\uprho}_1{\mathrm{lnNFOS}}_{\mathrm{t}}+{\uprho}_2{\mathrm{lnRGDP}}_{\mathrm{t}}+{\uprho}_3{\mathrm{lnURB}}_{\mathrm{t}}+{\uprho}_4{\mathrm{lnOPEN}}_{\mathrm{t}}+{\upmu}_{\mathrm{t}} $$

(3)

where δ₀ and ρ₀ are the intercepts, δ_iand ρ_i (i = 1, …, 4) are the elasticity parameters to be estimated and ϑ_tand μ_t are the residual error-terms for the respective models. The variables FOS and NFOS refer to the per capita fossil and nonfossil fuel consumption levels.⁵ Due to fossil fuels being relatively environment unfriendly in nature, the elasticity parameter δ₁ can be expected to depict a positive sign (Li et al. 2019). In contrast, nonfossil fuels are comparatively cleaner and therefore environmentally friendly. Thus, greater use of these energy resources can be linked to lower environmental damage. Consequently, the elasticity parameter ρ₁can be expected to be negative (Destek and Sinha 2020).

For further robustness check of the findings, the impacts of natural gas and hydroelectricity consumptions on the CFP of Bangladesh are analyzed. The relevance behind the selection of these energy types is based on the understanding that natural gas is a fossil fuel that accounts for the largest share of Bangladesh’s total electricity outputs. On the other hand, the hydroelectricity consumption data is the only renewable energy (nonfossil fuel) consumption data that is available in the Bangladesh context. The underlying models can be specified as:

$$ {\mathrm{lnCFP}}_{\mathrm{t}}={\varphi}_0+{\varphi}_1{\mathrm{lnNGC}}_{\mathrm{t}}+{\varphi}_2{\mathrm{lnRGDP}}_{\mathrm{t}}+{\varphi}_3{\mathrm{lnURB}}_{\mathrm{t}}+{\varphi}_4{\mathrm{lnOPEN}}_{\mathrm{t}}+{\tau}_{\mathrm{t}} $$

(4)

$$ {\mathrm{lnCFP}}_{\mathrm{t}}={\omega}_0+{\omega}_1{\mathrm{lnHEC}}_{\mathrm{t}}+{\omega}_2{\mathrm{lnRGDP}}_{\mathrm{t}}+{\omega}_3{\mathrm{lnURB}}_{\mathrm{t}}+{\omega}_4{\mathrm{lnOPEN}}_{\mathrm{t}}+{\theta}_{\mathrm{t}} $$

(5)

where φ₀ and ω₀ are the intercepts, φ_i and ω_i (i = 1, …, 4) are the elasticity parameters to be estimated and τ_t and θ_t are the residual error-terms for the respective models. The variables NGC and HEC abbreviate for the per capita natural gas and hydroelectricity consumption levels in Bangladesh. The signs of the elasticity parameters φ₁and ω₁can be expected to be identical to the signs of the elasticity parameters δ₁and ρ₁ in models (2) and (3), respectively.

Annual data from 1975 to 2016 is used to perform the econometric analysis. All the variables have been transformed into their natural logarithms for the ease of estimating the conditional EFP elasticities and also to reduce the sharpness of the data to generate consistent estimates. Table 1 reports the descriptions and data sources of the variables.

Table 1 Description of variables and data sources

Table 2 provides the descriptive statistics and the correlative matrix concerning the aforementioned variables. It is evident from the descriptive statistics that all the variables are platykurtic due to their respective value of kurtosis being less than 3. The CFP, aggregate energy consumption, real GDP and trade openness are positively skewed while fossil fuel consumption, nonfossil fuel consumption, natural gas consumption, hydroelectricity consumption, and urbanization are negatively skewed. The correlation matrix denotes strong correlations between the variables.

Table 2 The descriptive statistics and correlative matrix

Methodology

The econometric analysis begins by checking the stationarity and cointegration properties of the variables. These are followed by the regression analysis to predict the long-run elasticities. Finally, the causality analysis is conducted to assess the direction of causality among the variables.

Unit root analysis

Identifying the order of integration among the variables included within an empirical model is a key to choosing the appropriate regression and causality estimators. Thus, the econometric analysis commences by checking for unit roots in the data. This approach denotes whether the variables are mean-reverting or not. The existence of unit root in the data leads to spurious elasticity estimates. Although the first generation unit root tests were popularly tapped in the existing literature, these primitive methods fall short to account for the SB in the data, thus, resulting in biased estimates of the stationarity properties (Perron 1990). To address this limitation, this paper employs the unit root estimation technique proposed by Perron and Vogelsang (1992). The Perron-Vogelsang (PV) test can accommodate one endogenous SB in the data. This paper considers the Additive Outliers (AO) model within the PV approach which accounts for a sudden shift in the mean of the data. The null hypothesis of nonstationarity under the AO model, as suggested by Perron and Vogelsang (1992), can be specified as:

$$ {\mathrm{y}}_{\mathrm{t}}=\upbeta \mathrm{D}{\left(\mathrm{TB}\right)}_{\mathrm{t}}+{\mathrm{y}}_{\mathrm{t}-1}+{\mathrm{e}}_{\mathrm{t}},\mathrm{t}-2,\dots, \mathrm{T} $$

(6)

where for a given series y at the break year T_b ranging from 1 to T (i.e., 1< T_b< T). D (BY)_t = 1 if t = T_b + 1 and 0 if T_b is less than 1. The error sequence e_t is a finite-order stationary and invertible Autoregressive Moving Average (ARMA(p,q)) process. The alternative hypothesis of stationarity can be shown as:

$$ {\mathrm{y}}_{\mathrm{t}}=\partial +{\upbeta \mathrm{DU}}_{\mathrm{t}}+{\mathrm{v}}_{\mathrm{t}} $$

(7)

where v_t is also a finite-order stationary and invertible ARMA(p+1,q) process and DU_t = 1 if t > T_b and DU_t = 0 if t is less than or equal to T_b. Hence, the statistical significance of the t-statistic for testing the null hypothesis would reject the null hypothesis to confirm the stationarity of the series at its level of first difference form. Although the AO model considers a sudden shift in the mean value of the series, this paper also uses the Innovational Outliers (IO) model which considers the shift in the mean in a gradual manner.

However, since Bangladesh has experienced several episodes of macroeconomic shocks including the Asian financial crisis in 1997, it is important to accommodate more than one structural break within the unit root estimation process. Thus, the Narayan-Popp (NP) method (Narayan and Popp 2013) is also considered in this paper. This technique is superior to the PV method since it accounts for two SBs in the data. Narayan and Popp (2013) recommended two models, based on the location of the structural breaks, to estimate the stationarity properties of the variables. In the first model, the structural breaks are considered to be in the intercept only while the second model assumes the structural breaks to be both in the intercept and in the slope of the trend function (Narayan and Popp 2013). The test statistics under both the PV and NP unit root tests are predicted under the null hypothesis of nonstationarity of the series against the alternative hypothesis of stationarity. The unit root analysis is followed by the analysis of the cointegrating properties of the variables of concern.

Cointegration analysis

Once the order of integration among the variables is ascertained from the unit root analyzes, it is important to evaluate the cointegrating relationship between them. The cointegration exercise helps to understand whether there is any long-run association between the variables included within the respective models. Moreover, cointegration among the variables is also said to be a prerequisite to estimating the long-run causal associations. While the traditional time-series cointegration estimation techniques developed by Engle and Granger (1987), Johansen (1991), and Johansen and Juselius (1990) have been widely tapped in the existing studies, these methods are incapable of accommodating the SB issues in the data. All these techniques are based on the incorrect assumption of the cointegrating associations being time-invariant. However, SB in the data nullifies this erroneous assumption which makes their application inappropriate. Thus, following Murshed and Alam (2021), this paper uses Maki (2012) cointegration method which can account for as many as five SB in the data. However, considering the short time period of analysis in this study, the maximum number of SB has been limited to two.

Under the Maki (2012) approach, there are four specific models: A (considering the breaks in the intercept only), B (considering the breaks in the intercepts and coefficients without the trend, C (considering the breaks only in the intercept and coefficients but the model is assumed to have a trend), and D (considering the breaks in the intercepts, coefficients and trend). These models can be specified as:

$$ \mathrm{Model}\ \mathrm{A}:\kern1.5em {\mathrm{y}}_{\mathrm{t}}=\partial +{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\partial}_{\mathrm{i}}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+{\upbeta}^{\prime }{\mathrm{x}}_{\mathrm{t}}+{\mathrm{u}}_{\mathrm{t}} $$

(8)

$$ \mathrm{Model}\ \mathrm{B}:\kern1.5em {\mathrm{y}}_{\mathrm{t}}=\partial +{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\partial}_{\mathrm{i}}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+{\upbeta}^{\prime }{\mathrm{x}}_{\mathrm{t}}+{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\upbeta}_{\mathrm{i}}^{\prime }{\mathrm{x}}_{\mathrm{t}}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{t}} $$

(9)

$$ \mathrm{Model}\ \mathrm{C}:\kern1.5em \mathrm{v}{\mathrm{y}}_{\mathrm{t}}=\partial +{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\partial}_{\mathrm{i}}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+\upgamma \mathrm{t}+{\upbeta}^{\prime }{\mathrm{x}}_{\mathrm{t}}+{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\upbeta}_{\mathrm{i}}^{\prime }{\mathrm{x}}_{\mathrm{t}}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{t}} $$

(10)

$$ \mathrm{Model}\ \mathrm{D}:\kern1.5em {\mathrm{y}}_{\mathrm{t}}=\partial +{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\partial}_{\mathrm{i}}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+\upgamma \mathrm{t}+{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\upgamma}_{\mathrm{i}}\mathrm{t}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+{\upbeta}^{\prime }{\mathrm{x}}_{\mathrm{t}}+{\sum}_{\mathrm{i}=1}^{\mathrm{m}}{\upbeta}_{\mathrm{i}}^{\prime }{\mathrm{x}}_{\mathrm{t}}{\mathrm{D}}_{\mathrm{i},\mathrm{t}}+{\mathrm{u}}_{\mathrm{t}} $$

(11)

where y is the dependent variable and x is a vector of the independent variables. t denotes the time period, m denotes the highest number of breaks in the model which can take a maximum value of 5 and u denotes the error term. D_i,t is the dummy variable that is used to signal the presence of a break in the data. D_i,t takes a value of 1, denoting the existence of the SBs at specific break years T_bi (i=1, …, m), if t_i are after the break years T_bi (i.e., t_i > T_bi) and a value of 0, denoting no SB at the specific break years T_bi (i=1, …, m), if t_i are before the break years T_bi (i.e., if t_i < T_bi). Statistical significance of the test statistics in each of the four models (A, B, C, and D) rejects the null hypothesis of no cointegration to confirm the presence of cointegrating associations between the variables in the respective models.

Identification of the SB is even more justified in the sense that Bangladesh has achieved its independence in 1971, following the end to a grueling 9 months of war against Pakistan. Thus, just after independence, the economy of Bangladesh could not take off in a robust manner which could have possibly influence the energy consumption-environmental quality nexus that is being explored. Hence, to account for these issues, the possible SB years identified from the Maki (2012) cointegration analysis are used as break year dummies and augmented into the respective regression models to estimate the elasticities taking the presence of the SB issues into account.

Regression analysis

Following Kirikkaleli et al. (2021), the long-run elasticities are estimated using the conventionally used time-series Fully-Modified Ordinary Least Squares (FMOLS) estimator of Phillips and Hansen (1990). The FMOLS estimation technique, proposed by Phillips and Hansen (1990), is grounded on a nonparametric approach to regression analysis which provides estimates accounting for potential endogeneity and serial correlation issues in the data. Moreover, the FMOLS estimator is ideally suited for estimations involving series that are integrated at their first differences (Amarawickrama and Hunt 2008). The FMOLS estimator $\left({\hat{\partial}}_{FMOLS}\right)$ can be specified as:

$$ \left({\hat{\partial}}_{\mathrm{FMOLS}}\right)={\mathrm{N}}^{-1}{\sum}_{\mathrm{i}=1}^{\mathrm{N}}\hat{\partial_{\mathrm{FMOLS},\mathrm{i}}} $$

(12)

The corresponding t-statistic from the FMOLS estimator can be expressed as:

$$ {\mathrm{t}}_{{\hat{\partial}}_{\mathrm{FMOLS}}}={\mathrm{N}}^{-1/2}{\sum}_{\mathrm{i}=1}^{\mathrm{N}}{\mathrm{t}}_{{\hat{\partial}}_{\mathrm{FMOLS}}} $$

(13)

For robustness check, the Dynamic Ordinary Least Squares (DOLS) estimator of Stock and Watson (1993) is also applied.

Causality analysis

Following Murshed (2021c), the causal associations between the variables are evaluated using the Hacker-Hatemi-J (HH) causality estimation technique proposed by Hacker and Hatemi-J (2012). This method is modified from the bootstrapped causality test proposed by Hacker and Hatemi-J (2006). According to Hacker and Hatemi-J (2006), bootstrapping the distribution reduces the concerns from small sample size distortions associated with the conventional Wald test of cauasality introduced by Toda and Yamamoto (1995), irrespective of the presence of autoregressive conditional heteroscedasticity effects within the model. The modification of the modified Wald test statistic of Hacker and Hatemi-J (2006) is done by endogenizing the optimal lag selection criterion which tends to reduce the small sample size distortions further. The HH technique uses a Vector Autoregression (VAR) model to calculate the modified Wald test statistics under the null hypothesis of no causality between the dependent and the independent variables against the alternative hypothesis of otherwise. The bootstrapping involves two stages: firstly, estimating the optimal lag structure and secondly predicting the Wald statistic for investing the Granger causality. The VAR model of order k can be specified as:

$$ {\mathrm{y}}_{\mathrm{t}}={\upbeta}_0+{\upbeta}_1{\mathrm{y}}_{\mathrm{t}-1}+\dots +{\upbeta}_{\mathrm{k}}{\mathrm{y}}_{\mathrm{t}-\mathrm{k}}+{\mathrm{u}}_{\mathrm{t}} $$

(14)

where y_t, β₀, and u_t are vectors with dimension n x 1 and B_i (i > 0) is a parameter matric with a dimension of n x n. The error term u_t has no expected value and presumed to be independent and identically distributed with a nonsingular covariance matrix (Hacker and Hatemi-J 2012). In the first stage of the bootstrapping approach under the HH approach, Eq. (14) is estimated without imposing any restriction in terms of the noncausality null hypothesis. The predicted value, y*, can be given by:

$$ {{\mathrm{y}}^{\ast}}_{\mathrm{t}}={\hat{\upbeta}}_0+{\hat{\upbeta}}_1{\mathrm{y}}_{\mathrm{t}-1}+\dots +{\hat{\upbeta}}_{\mathrm{k}}{\mathrm{y}}_{\mathrm{t}-\mathrm{k}}+{\hat{\mathrm{u}\ast}}_{\mathrm{t}} $$

(15)

where ${\hat{u\ast}}_t$ is a vector of bootstrapped error terms and t (t=1, …, T) is the time period. The set of T bootstrapped error term vectors is estimated by drawing randomly with replacement from the vectors of the modified residual to ensure that the mean value of the bootstrapped error term vectors is a zero vector. The modified residuals are the raw residuals modified via leverages within the HH approach. This modification is ideal to deal with heteroscedasticity issues within the model and also to account for the Autoregressive Conditional Heteroskedasticity (ARCH) effects (Hacker and Hatemi-J 2012). The bootstrapping mechanism is repeated M times to produce a Wald statistic each time which is based on the TY (2005) methodology. The resulting set of bootstrapped Wald statistics is termed as the bootstrapped Wald distribution which is then used to evaluate the causal properties of the pair of variables. In the case of the predicted Wald—statistic being greater than the corresponding bootstrapped critical value then causality stemming from the independent to the dependent variable is affirmed via rejection of the null hypothesis of noncausality.

Results and discussions

This section begins by reporting the findings from the unit root analysis which is reported in Table 3. The from both the PV and NP tests reveal that the null hypothesis on nonstationarity for all the variables cannot be rejected at their level forms but can be rejected at their first differences. The statistical significance of the predicted test statistics affirms this claim. These findings implicate that all the variables are mean-reverting whereby the possibility of conducting spurious regression analyzes is nullified. The unit root analysis is followed by the cointegration analysis.

Table 3 The unit root test results

The results from the Maki (2012) cointegration analysis are reported in Table 4. The statistical significances of the estimated test statistics at 1% reject the null hypothesis of noncointegration to confirm the long-run associations between the variables in all the three regression models. Hence, it can be claimed that there are long-run relationships between per capita CFP, per capita energy consumption (in aggregate and also in disaggregated forms of fossil, nonfossil fuel, natural gas and hydroelectricity consumptions), economic growth, urbanization, and trade openness. Besides, the two SB for each of the models are also ascertained. In order to account for the SB issues, the SB identified from Model A under the Maki (2012) cointegration analysis are used to construct dummy variables which are then included into the respective models. The confirmation of long-run associations between the variables fulfills the pre-requisite to estimating the long-run elasticities.

Table 4 The Maki (2012) cointegration test results

Table 5 reports the long-run elasticity estimates from the FMOLS regression analysis. The elasticity estimates reveal that higher aggregate energy consumption levels tend to boost the CFP of Bangladesh in the long-run. A 1% rise in the per capita total energy consumption levels is predicted to increase the per capita CFP by 0.19%–0.21%, on average, ceteris paribus. Hence, it can be said that although energy use in Bangladesh has stimulated the growth of the economy, it has deteriorated the environment in Bangladesh. This finding is justified given the predominant fossil fuel dependency of Bangladesh whereby fossil fuels account for a major proportion of the nation’s primary energy and electricity consumption levels. A similar positive correlation between overall energy consumption and EFP was reported in the studies by Destek and Sarkodie (2019) and Baloch et al. (2019) for 11 newly industrialized economies and 59 BRI nations including Bangladesh. Although the aggregate energy consumption levels are found to be positively associated with the CFP, heterogeneous impacts are ascertained at the disaggregated per capita energy consumption levels. The elasticity estimates reveal that higher nonfossil fuel consumption results in environmental improvement by reducing the CFP. In contrast, higher fossil fuel consumption aggravates environmental quality by increasing the CFP. A 1% rise in the per capita fossil and nonfossil fuel consumption levels are predicted to increase and decrease the per capita CFP by 0.21%–0.25% and 0.18%–0.19%, respectively, ceteris paribus. Since Bangladesh has traditionally been overwhelmingly reliant on fossil fuels to meet its energy demand, these contrasting findings statistically certify the factors that have contributed to the persistent deterioration of the nation's environmental quality. Hence, the government’s recent decision to introduce coal into the national energy mix could further jeopardize the prospects of ensuring environmental sustainability in Bangladesh. Rather, it is ideal for the nation to integrate renewable energy resources into the energy system to gradually phase out fossil fuel dependency. These findings are parallel to the findings reported by Ulucak et al. (2020) for the OECD countries. The authors asserted that higher consumption of nonrenewable energy was responsible for higher CFP levels while and increasing renewable energy reduced the CFP of these nations.

Table 5 The Long-run elasticities from the FMOLS regression analysis

For a robustness check of the long-run elasticity estimates, the impacts of natural gas and hydroelectricity consumptions on the CFP are also explored. The corresponding elasticity estimates show that a 1% rise in the per capita natural gas consumption level in Bangladesh stimulates a rise in the per capita CFP by 0.15%–0.16%, on average, ceteris paribus. On the other hand, a 1% rise in the per capita hydroelectricity consumption levels is found to reduce the 0.02%–0.03%, on average, ceteris paribus. Hence, these findings corroborate the aforementioned findings of fossil fuel consumption deteriorating the environmental quality of Bangladesh while nonfossil fuel consumption improving it in the long-run. However, it can be seen that the marginal impacts of hydroelectricity consumption on the CFP are extremely low. This could be because Bangladesh merely generates a little more than 1% of its total electricity outputs using renewable resources. Consequently, such low levels of hydroelectricity are justifiably unable to exert notable impacts in curbing the CFP.

Among the other key determinants of CFP, the elasticity estimates show that economic growth in Bangladesh is not conducive to ensuring environmental sustainability. These positive signs of the corresponding elasticities imply that as the per capita real GDP level of Bangladesh increases, the long-run per capita CFP tend to go up as well. Hence, it can be said that the nation has traditionally traded off higher economic growth for lower environmental quality. This can also be explained in terms of Bangladesh’s energy consumption trends which have predominantly been spearheaded by fossil fuels including natural gas, petroleum oils and coal. Besides, the environmental regulations in Bangladesh, much like the other similar developing nations, are not strong enough to ensure the protection of the environmental attributes. As a result, the production and consumption processes in Bangladesh have not been sustainable which, in turn, have contributed to the worsening of the environmental quality. Similar findings were also featured in the study by Hassan et al. (2019) in the context of another developing South Asian economy Pakistan. The findings in that study showed that a 1% rise in the real GDP level of Pakistan caused a 0.41% rise in the nation's EFP in the long-run. Since both Bangladesh and Pakistan share similar economic and energy consumption structures, it can be said that both these nations are yet to effectively safeguard their respective environment.

On the other hand, the elasticity estimates reported in Table 5 shows that the urbanization rate is positively correlated to the per capita CFP in Bangladesh. However, the elasticity estimates are not statistically significant which implies that urbanization cannot explain the variations in Bangladesh’s long-run per capita EFP. This finding contradicts the results asserted in the study by Ulucak and Khan (2020) in which the authors claimed that urbanization mitigates the EFP in the BRICS countries. Finally, the elasticity estimates indicate that international trade is also detrimental to Bangladesh’s environmental sustainability. The positive signs of the elasticity parameters affirm this claim. Since Bangladesh pursues an export-led growth strategy, this finding imposes ominous concerns over the nation’s overall quality of the environment. Once again, it is evident that the weak environmental regulations in Bangladesh have enabled the nation to specialize in pollution-intensive production of export items. The finding is similar to the conclusions put forward by Aydin and Turan (2020) for South Africa. However, the authors also found evidence of higher openness to trade resulting in lower EFP in India.

Notes: The optimal lag selection is based on AIC; The standard errors are reported within the parentheses; SB1 and SB2 refer to the two SB dummy variables identified from model A under the Maki (2012) cointegration analysis; ***, **, and * denote statistical significance at 1%, 5%, and 10% levels, respectively

The robustness of the long-run elasticity estimates is checked using the DOLS estimator. The results from the DOLS analysis are presented in Table 6. The similarity in respect of the predicted signs of the DOLS and FMOLS elasticity estimates affirm the robustness of the regression outcomes across the alternative methods.

Table 6 Robustness analysis using the DOLS method

Notes: The optimal lag selection is based on AIC; The standard errors are reported within the parentheses; SB1 and SB2 refer to the two SB dummy variables identified from model A under the Maki (2012) cointegration analysis; ***, **, and * denote statistical significance at 1%, 5%, and 10% levels, respectively

Table 7 presents the causality findings from the HH bootstrapped causality test. It is apparent that total energy consumption in Bangladesh affects the nation’s CFP. The estimates indicate a unidirectional causal association stemming from aggregate energy consumption to CFP, without the feedback. Similar unidirectional causation from fossil fuel consumption to CFP is also ascertained. On the other hand, a bidirectional causal association between nonfossil fuel consumption and CFP is confirmed from the HH causality analysis. This feedback causal association suggests that environmental degradation in the form of rising CFP levels are likely to trigger awareness regarding the need for ensuring environmental sustainability within Bangladesh. This, in turn, can be presumed to enhance nonfossil fuel consumption and facilitate the CET phenomenon in Bangladesh. Furthermore, as far as the natural gas and hydroelectricity consumptions are concerned, the causality estimates reveal unidirectional causations running from both natural gas and hydroelectricity consumption to Bangladesh’s CFP. Hence, keeping into consideration the corresponding elasticity estimates, these causality findings indicate that energy consumption plays a vital role in influencing Bangladesh’s CFP. Thus, it is pertinent for the nation to reduce the monotonic fossil fuel dependency and gradually undergo CET to reduce the CFP and ensure environmental sustainability. The finding of the unidirectional causality running from total energy consumption to CFP is parallel to the findings reported by Destek and Sarkodie (2019) for 11 newly industrialized economies.

Table 7 Hacker and Hatemi-J bootstrapped causality test results

Besides, the causality estimates reported in Table 7 confirm the existence of a unidirectional causal relationship stemming from the per capita real GDP and the CFP levels of Bangladesh. Hence, this finding provides support to the parallel elasticity estimates and certifies that the growth of the Bangladesh economy has resulted in the worsening of the environment. This traditional trade-off between economic and environmental welfares needs to be addressed in order to ensure complementarity between economic and environmental development in Bangladesh. Similarly, a unidirectional causal influence of economic growth on the EFP for the case of European nations was also highlighted in the study by Alola et al. (2019). On the other hand, no causal association between urbanization and CFP in Bangladesh could be ascertained from the causality analysis. However, a unidirectional causality running from trade openness to CFP is evidenced. Therefore, this finding supports the parallel elasticity estimate to highlight the adverse impacts of international trade on Bangladesh’s environmental quality. As a result, this particular finding imposes doubt on the credibility of the international trade policies of Bangladesh which seem to be not aligned with the environmental sustainability agenda of the government. As opposed to the unidirectional causal nexus between trade openness and CFP, Lu (2020) found bidirectional causation between trade openness and EFP in the context of 13 Asian economies including Bangladesh. Figure 4 presents a summary of the causal associations ascertained from the HH causality test.

Figure 4:

Illustration of the causality findings Note: ↔ indicates bidirectional causality and → indicates unidirectional causality

Conclusion and policy implications

In the contemporary era, achieving a low-carbon economy has become a global agenda. In line with this notion, the majority of the world economies have ratified the Paris Agreement and have committed to attaining carbon-neutrality at the earliest. Besides, the global economies have also agreed to facilitate the objectives of the United Nations SDG agenda, especially focusing on ensuring environmental sustainability across the globe. Achieving a low-carbon economy is particularly important for developing nations that are predominantly fossil-fuel dependent, to which Bangladesh is no exception. Therefore, this study specifically aimed at evaluating the environmental impacts associated with energy consumption and other key macroeconomic variables in the context of Bangladesh over the 1975–2016 period. As opposed to the conventional practice of using CO₂ emissions to proxy environmental quality, this study makes a novel attempt to use the CFP to measure environmental welfare. The econometric analyzes involved the application of methods that are suitable for handling the SB issues in the data. The overall findings from empirical exercises revealed that aggregate energy consumption, fossil fuel consumption, and natural gas consumption boost the CFP of Bangladesh. In contrast, nonfossil fuel consumption and hydroelectricity consumption are found to abate the CFP figures. Besides, economic growth and international trade are also evidenced to increase the CFP in Bangladesh. Furthermore, the causality analysis confirmed the presence of unidirectional causalities stemming from total energy consumption, fossil fuel consumption, natural gas consumption, hydroelectricity consumption, economic growth, and international trade to the CFP. On the other hand, nonfossil fuel consumption is found to be bidirectionally associated with the CFP. In line with these aforementioned findings, it can be said that reducing fossil fuel dependency by promoting relatively cleaner energy use is key to achieve carbon-neutrality in Bangladesh. Therefore, the overall findings from this study can be expected to enable Bangladesh to comply with its commitments concerning the attainment of the environmental welfare targets enlisted under the Paris Agreement and the United Nations SDG declarations.

Several policies can be put forward in the context of Bangladesh’s future strategies to achieve a low-carbon economy. Firstly, since the results indicated that the traditional fossil fuel dependency of Bangladesh is one of the major factors responsible for the surge in the nation’s CFP figures during the period of this study, it is ideal for the government to diversify the national energy mix. This diversification should ensure the replacement of conventionally used fossil fuels with relatively cleaner alternative fuels. Thus, CET is a must for the nation to attain carbon-neutrality in the future. In this regard, the government must invest in renewable energy technology development in order to elevate the share of renewables in the total final energy consumption and electricity output figures of Bangladesh. Besides, energy infrastructural development should also be prioritized to further facilitate the integration of renewable energy into the domestic energy system. Simultaneously, Bangladesh should also try to attract foreign direct investments toward its renewable energy sector which can be expected to bridge the nation’s technological constraints which have traditionally inhibited the CET phenomenon in Bangladesh. Furthermore, the government should also adopt appropriate international trade policies to make way for Bangladesh to import renewable electricity from neighboring South Asian nations like Nepal and Bhutan. Such cross-border intra-regional flows of renewable energy into Bangladesh can be anticipated to reduce the nation’s CFP figures further.

Secondly, since the findings also affirmed a trade-off between economic growth and environmental degradation in Bangladesh, it is pertinent for the government to infuse the carbon-neutrality objective within the national economic growth agenda. Consequently, complementarity between economic and environmental development can be ensured. Once again, CET can be said to be central for simultaneously achieve economic and environmental welfares. Therefore, it is recommended that the Bangladesh government would enact stringent environmental policies which would be effective in inhibiting fossil fuel use within the production processes. At the same time, the government should also incentivize renewable energy utilization within the production processes by extending low-interest loans to private sector projects that are directly linked with renewable energy development in Bangladesh. This is extremely important because the involvement of the private sector in Bangladesh’s energy sector is minimal. Thus, encouraging private sector participation in setting up renewable power plants would not only contribute to the nation’s economic growth but would also enable the nation to gradually become a low-carbon economy. Lastly, the government is also required to revisit the existing international trade policies to counter the adverse environmental impacts of international trade in Bangladesh. In this regard, imposing barriers to the trade of carbon emission-intensive goods and services can play a vital role. Simultaneously, the stringency of the environmental protection laws should be enhanced so that the foreign countries do not exploit the weak environmental regulations of Bangladesh to relocate their dirty industries to Bangladesh. Furthermore, it is recommended that Bangladesh includes renewable energy in its import basket which can effectively facilitate the CET phenomenon to help the nation achieve its carbon-neutrality objectives. More importantly, along with restricting the trade of carbon-intensive goods and services, it is critically important for Bangladesh to lower the trade barriers that have traditionally impeded the nation from trading with its South Asian neighbors, especially in respect of facilitating intra-regional trade of renewable energy resources. Such amendments in the existing trade policies can be envisioned to not only reduce the adverse environmental consequences associated with international trade participation by Bangladesh, but it would also help the nation to let go of its traditional fossil fuel-dependency tag.

Data constraint was the major limitation faced in this paper. Due to the unavailability of data, the impacts of different nonfossil fuels on the CFP of Bangladesh could not be investigated. It would have been ideal to further disaggregate the fuel consumption data for unearthing robust policy implications concerning environmental sustainability in Bangladesh. As part of the future scope of research, this paper can be extended by controlling for institutional quality since the poor quality of institutions is one of the other key factors responsible for environmental degradation in Bangladesh.