Mitigating Transport-Based CO₂ Emissions in Landlocked Countries: The Role of Economic Growth, Trade Openness, Freight Transportation and Renewable Energy Consumption

Abstract

The transportation sector plays a pivotal role in economic development but is also a major contributor to environmental degradation due to its reliance on fossil fuels. This study explores the relationship between transport-related CO₂ emissions, economic growth, road and rail freight transport, industry, trade openness, fossil fuel consumption, financial development, and renewable energy in ten landlocked countries from 1990 to 2022. Using panel cointegration tests and PMG-ARDL techniques, the findings reveal a bidirectional causality between CO₂ emissions, road freight, financial development, and industry. Road freight transport significantly boosts economic growth but also intensifies emissions, while renewable energy effectively mitigates transport-related CO₂. The results emphasize the need for policymakers to balance economic advancement with sustainable energy and emission reduction strategies. Achieving economic-energy sustainability is essential for fostering a green and clean environment without compromising growth.

1. Introduction

In a competitive global setting, nations utilize their natural resources to achieve higher levels of economic growth. However, they often overlook the environmental consequences of their activities, which can lead to degradation. The relationship between the transport sector, economic development, industrialization, trade openness, and increasing fossil fuel consumption, and their combined impact on the environment, has been the subject of extensive debates and academic studies in LLCs for quite some time. This dynamic relationship, which involves the economic activity, the transport sector, and the connection between reducing fossil fuel consumption and adopting renewable energies to minimize environmental degradation, presents conflicting and paradoxical objectives for decision-makers. The transportation industry primarily relies on fossil fuels, particularly oil and gas, which release significant quantities of GHGs and contribute to accelerating climate change [1,2]. The negative effects of environmental degradation are becoming increasingly evident, especially in the form of ambient air pollution. It is crucial to recognize that the power, industry, and transport sectors are the main contributors to fossil fuel-related carbon dioxide emissions globally. While there has been extensive analysis of the factors influencing CO₂ emissions and emission intensities in the industry and power sectors across numerous countries, there is a notable lack of examination regarding emissions and emission intensities in the transport sector. To ensure a sustainable future for the transport sector, it is crucial to analyze the correlation between transportation, economic growth, and the negative effects on the environment, particularly as measured by carbon dioxide emissions.

Currently, research on the relationships between logistics operations, financial development, fossil fuel consumption, renewable energy consumption, and industrialization in LLCs is limited, especially regarding their impact on transport-related carbon dioxide emissions. Understanding how transportation contributes to environmental degradation is essential for assessing the effectiveness of current energy strategies in this sector. If required, appropriate measures should be taken to address existing issues [3]. Transportation is a vital sector for sustainability, significantly influencing the environment, economy, and society. Over the last 30 years, there has been increasing attention from policymakers, the transport industry, and academia on the concept of sustainable transportation, which is grounded in the idea of sustainable development. In this context, it is important to examine the negative impacts of transportation on sustainable development so that future policymakers can consider sustainable fuel alternatives for the transportation sector. Prioritizing the promotion of clean energy and a reduction in pollutant emissions is essential to fostering environmental protection. There is a strong correlation between the transportation sector, economic growth, energy consumption, CO₂ emissions, and freight transport. To achieve sustainable transportation goals, it is necessary to lessen the impact of environmental change. Ensuring a clean environment for future generations presents substantial challenges for many countries worldwide. The term “landlockedness” refers to the geographical condition of a country that lacks direct access to the sea. Currently, 44 countries are classified as landlocked, with no coastal access. Among these, the UN identifies 32 as landlocked, which fall under the low- and middle-income categories according to the World Bank’s classification. These countries face complex and interconnected economic, social, and environmental issues, necessitating the implementation of effective development policies to address these challenges [4].

Although the region has abundant energy supplies, its energy consumption remains a significant cause of pollution. This study significantly advances the current understanding by addressing a critical, yet underexplored, dimension of environmental economics: the specific determinants of transport-related CO₂ emissions within the unique and challenging context of LLCs [5,6]. While a considerable body of literature examines the broader nexus of economic growth, energy consumption, and emissions, there remains a conspicuous paucity of research that holistically investigates the interplay of trade openness, the distinct roles of road versus rail freight transport, financial development, industrial structure, and the pivotal contribution of renewable energy consumption in mitigating these emissions specifically for LLCs [5,6,7]. Prior research has often overlooked the nuanced logistical and economic vulnerabilities inherent to these nations or has failed to concurrently analyze this comprehensive suite of variables using robust and contemporary panel data methodologies. This research diverges from previous studies on the EKC, which typically evaluate aggregate emissions or depend primarily on economic growth as the main explanatory variable. Instead, it employs a sector-specific approach, concentrating specifically on the transportation sector, and incorporates modal freight disaggregation, distinguishing between road and rail transport. This facilitates the assessment of both the magnitude of emissions and the effectiveness and sustainability of various freight systems, which holds particular significance for landlocked nations that depend extensively on overland transportation. Moreover, the integration of PMG-ARDL, with impulse response analysis and variance decomposition, facilitates the identification of dynamic and directional effects, surpassing conventional static estimations. Consequently, this paper fills an important void by providing a focused empirical analysis of ten LLCs from 1990 to 2022, employing advanced panel cointegration tests and PMG-ARDL techniques. The novelty of this research lies not only in its geographical and temporal scope but also in its capacity to disentangle complex and bidirectional causal relationships and to offer granular insights into how factors, such as renewable energy adoption, can specifically curb transport sector emissions in these often-marginalized economies [6]. The findings are therefore of paramount importance for policymakers, offering a data-driven foundation for crafting targeted strategies that can promote sustainable transport, foster green economic growth, and enhance energy security, thereby guiding LLCs towards a more resilient and environmentally sound development trajectory. The remainder of the article is arranged as follows: Section 2 reviews some background studies. Section 3 shows the proposed methodology. Section 4 presents and discusses the results. Section 5 proposes the findings and the policy implications.

2. Literature Review

Human activities, including economic activities like manufacturing goods, providing services, and utilizing infrastructure for commercial operations and transport, contribute to environmental degradation [8]. Environmental regulations are established to safeguard individual welfare by monitoring these economic and social activities. Previous research on environmental standards has focused on evaluating the distinct relationships between transportation, industrialization, and environmental issues. These relationships are often discussed in relation to CO₂ emissions. Below is an overview of the literature that highlights the connections between the variables of interest.

2.1. Nexus Between Financial Development, Energy Consumption, Economic Growth, and Environmental Degradation

The interplay among financial development, energy consumption, economic growth, and environmental degradation has been extensively researched due to its implications for sustainable development. Numerous studies highlight energy consumption, particularly from fossil fuels, as a significant contributor to environmental degradation, resulting in increased CO₂ emissions and ecological harm [9,10]. Financial development plays a critical role in this nexus, serving a dual purpose. On the one hand, it facilitates investments in renewable energy and cleaner technologies, potentially mitigating environmental harm [11,12]. On the other hand, financial development in certain regions, such as sub-Saharan Africa and nations involved in the Belt and Road Initiative, has been linked to increased energy consumption and ecological footprints [13,14]. Empirical studies reveal diverse causal relationships among these variables. Gursoy-Haksevenler et al. [15] found a negative but statistically insignificant correlation between financial indicators and energy consumption. Similarly, Vidyarthi [16] observed a reciprocal long-term relationship between GDP and energy consumption in South Asia, along with a unidirectional relationship where CO₂ emissions influence both GDP and energy consumption. In OECD nations, the EKC hypothesis has been validated, with CO₂ emissions declining after reaching a certain stage of economic development [17]. Notably, findings by Ahmad et al. [18] using a VECM suggest that energy consumption, GDP, and CO₂ emissions are cointegrated in the long term, underscoring the enduring interdependence of these variables. Collectively, this body of research emphasizes that aligning financial development and economic growth with sustainable energy practices and robust environmental policies is essential for effectively mitigating environmental degradation.

2.2. Nexus Between Renewable Energy, Energy Consumption and Environmental Degradation

The global transition to clean energy, aimed at reducing CO₂ emissions, combating climate change, and promoting RE, has been extensively studied by Mohammed et al. [19]. In the EU, Bölük and Mert [20] reported that both RE and NREC increased CO₂ emissions, whereas Ben Jebli et al. [21] observed reductions in CO₂ emissions in 25 OECD countries due to RE. Similarly, Bilgili et al. [22] found that RE significantly decreased CO₂ emissions in 17 OECD countries over the period 1977–2010, and Zoundi [23] and Hu et al. [24] confirmed the positive impact of RE on emissions reduction in Sub-Saharan Africa and 25 developing countries, respectively. Sharif et al. [25], by using data from 74 countries, highlighted that EC increases emissions while RE reduces them. In the transportation sector, Raihan et al. [26] and Wang et al. [27] emphasized the benefits of RE, including enhanced energy security and reduced dependence on fossil fuels. Specific studies, such as those by Murshed et al. [28] in Argentina and Zahoor et al. [29] in China, confirmed that RE effectively lowers carbon dioxide emissions in this sector. Finally, the relationship between RE and economic growth (GDP) has been explored with Fan and Hao [1], identifying a stable and unidirectional effect of GDP on RE in 31 Chinese provinces over the period 2000–2015.

2.3. Nexus Between Trade Openness and Environmental Degradation

The relationship between trade openness and environmental quality has been a focal point in economic and environmental research. For instance, Gu et al. [30] identified a unidirectional causal relationship where foreign trade dependency influences CO₂ emissions. Contrarily, Akin [31] argued that trade openness decreases CO₂ emissions, establishing a unidirectional link from CO₂ to trade openness. Several studies have examined the regional impacts of trade openness. Hakimi and Hamdi [32] attributed negative environmental impacts in Morocco and Tunisia to trade openness. Similarly, Ling et al. [33] reported a positive correlation between trade openness and CO₂ emissions in the ASEAN-5 countries (Indonesia, Malaysia, the Philippines, Singapore, and Thailand) from 1995 to 2014, indicating a bidirectional relationship. Boutabba [34], in an analysis of India, found a negative association between trade openness and CO₂ emissions, highlighting a long-term equilibrium.

In Sri Lanka, Naranpanawa [35] concluded that while there is no long-term causality between trade openness and CO₂ emissions, short-term connections are evident. Meanwhile, Ertugrul et al. [36], focusing on ten developing countries (e.g., China, India, Brazil, and South Africa) from 1971 to 2011, identified cointegration among trade openness, energy consumption, real income, and CO₂ emissions, supporting the EKC hypothesis in several nations. In the United States, Dogan and Turkekul [37] demonstrated a reciprocal relationship between CO₂ emissions and trade openness from 1960 to 2010, emphasizing the importance of trade openness and urbanization in managing GDP and pollution. Zamil et al. [38] observed a positive relationship between trade openness, GDP per capita, and CO₂ emissions in Oman, urging policymakers to integrate environmental considerations into trade policies. Rahman et al. [39] explored South Asia over the period 1990–2017, finding a negative relationship between trade openness, CO₂ emissions, and economic growth, alongside unidirectional causality between these variables. Similarly, Yu et al. [40], studying CIS countries from 2000 to 2013, found a direct positive impact of trade openness on CO₂ emissions but a negative indirect effect via reduced per capita income. Lastly, Jun et al. [41], through wavelet-coherence analysis, demonstrated a dynamic relationship between trade openness and CO₂ emissions in China. Their findings highlighted the necessity for China to adopt strategies to mitigate pollution while pursuing trade liberalization.

2.4. Nexus Between Freight Transportation and Environmental Degradation

The relationship between transportation, energy consumption, economic growth, and environmental degradation has been a significant area of research, with findings highlighting various regional and sectoral dynamics. Adom et al. [42] examined the long-term correlation between road transport energy consumption and economic growth in six West African countries, identifying a unidirectional causal relationship from income to transport energy consumption using the Dumitrescu-Hurlin panel causality test. In Tunisia, Achour and Belloumi [43] revealed a positive link between GDP, transport value-added, infrastructure, and CO₂ emissions from 1971 to 2012, though no reverse causality was observed. Neves et al. [44], focusing on 15 OECD countries from 1995 to 2014, found that rail infrastructure investment could reduce fossil fuel usage and CO₂ emissions, highlighting the potential of sustainable transport policies. Further studies emphasized the environmental impact of freight transport. Song et al. [45] reported a 13.49% annual increase in traffic-related energy consumption in Shanghai from 2000 to 2010, while Murtishaw and Schipper [46] identified growing reliance on trucks in the U.S. as a driver of higher energy intensity. Linton et al. [47] forecast a 72% increase in rail transport energy use by 2050, with corresponding CO₂ emissions. Saidi and Hammami [48] emphasized the transportation sector’s role in global carbon emissions, while Shafique et al. [49] advocated for sustainable policies and technological advancements to mitigate these impacts. Collectively, these studies underscore the urgent need for environmentally conscious transportation strategies to balance economic growth with sustainability.

2.5. Nexus Between Industrialization and Environmental Degradation

The relationship between industrialization, energy consumption, economic growth, and environmental degradation has been extensively studied, with research consistently highlighting the role of industrialization in increasing carbon emissions. Carvalho et al. [50] emphasized that industrialization often prioritizes economic development at the expense of environmental sustainability. However, few studies have simultaneously examined the combined impact of industrialization, energy consumption, economic growth, and financial development on environmental degradation. Empirical evidence supports a positive link between industrialization and CO₂ emissions in various regions. For instance, Wang et al. [51] identified this association in Western developing countries, while Abou-Ali et al. [52] confirmed it for Arab regions. Rafiq et al. [53], using the STIRPAT model, demonstrated a positive relationship between industrialization and CO₂ emissions across 53 countries. Similar findings were reported by Nejat et al. [54] and Tian et al. [55] in China, Salim and Shafiei [56] in OECD countries, and Cherniwchan [57] across 157 countries. Shahbaz et al. [58] also established a link between industrialization and CO₂ emissions in Bangladesh. Several studies have explored the dynamic relationship between industrialization, energy consumption, and environmental degradation. Asumadu-Sarkodie and Owusu [59] found a long-term relationship between carbon emissions, energy use, industrialization, and financial development in Sri Lanka over the period 1971–2012, identifying a unidirectional causal relationship from energy consumption to carbon emissions and a bidirectional relationship between industrialization and energy use. Similarly, Asumadu-Sarkodie and Owusu [60] revealed a significant long-term link between industrialization and CO₂ emissions in Rwanda from 1965 to 2011, highlighting detrimental effects on health and air quality, and recommending environmentally friendly industrial policies. Appiah et al. [61] and Anwar and Elfaki [62] highlighted industrialization’s importance in promoting innovation and resource allocation but noted its environmental costs, particularly through increased energy consumption and CO₂ emissions. Nasrollahi et al. [63] underscored the positive impact of industrialization on economic growth, despite its challenges. As industrialization drives energy demand, its environmental implications become evident, necessitating policies that balance economic growth with environmental sustainability through technological advancements and structural reforms.

3. Data and Methodology

3.1. Data, Variables and Sources

Based on the availability of data, we chose a panel of ten LLCs from 1990 to 2022 using the World Bank (2024) [64] and the IEA (2024) [65] datasets. This particular period was chosen simply because the required data was not available for earlier periods. The selected countries included in the sample are Armenia, Austria, Azerbaijan, Belarus, Hungary, Kazakhstan, Luxembourg, Slovakia, Switzerland, and Uzbekistan. The selection is guided by the accessibility of data. Despite the differences in their level of economic development and geographical location (Europe and Asia), these countries exhibit notable common traits, including the lack of direct access to the sea, the significant dependence on land transport infrastructure, and the energy vulnerability resulting from their reliance on imports. The inclusion of these nations is motivated by the distinctive challenges faced by LLCs, such as reliance on transit nations for access to international markets, inadequate transportation infrastructure, complex customs and border procedures, and limited access to maritime trade routes, all of which significantly influence their economic and environmental dynamics.

The variables used in this analysis are defined as follows:

−

TCO₂: This variable is considered a dependent variable. This dimension is expressed as a percentage of total fuel combustion, which facilitates a focused examination of the environmental consequences attributable to the transportation sector. It is frequently neglected in research concerning landlocked nations.

−

ROFT: This variable quantifies the volume of goods transported via road, articulated in million ton-kilometers. Road transport exhibits a higher intensity of CO₂ emissions, primarily attributed to the predominant use of diesel-powered vehicles. The incorporation of this variable facilitates the evaluation of the environmental implications associated with short- and medium-haul freight transportation in landlocked regions.

−

RAFT: It pertains to the movement of freight via rail, quantified in million ton-kilometers. Rail transport generally exhibits greater energy efficiency and lower carbon intensity in comparison to road transport. The disaggregation of freight by mode, specifically comparing road and rail, facilitates the identification of modal shifts, which may serve as a viable mitigation strategy. Freight transport (ROFT and RAFT) exhibits a direct correlation with emissions levels. Road freight exhibits a considerably higher emission intensity, primarily attributed to its dependence on diesel fuel and the fragmented nature of logistics. In contrast, rail transport presents a more sustainable, lower-carbon alternative. The separation of freight modes facilitates the examination of potential shifts between different transportation methods.

−

GDP: It is quantified in constant US dollars and serves as a metric for assessing the magnitude of economic activity. Within the framework of the EKC, it is anticipated that GDP exhibits a non-linear correlation with CO₂ emissions. Initially, economic growth is associated with an increase in emissions, referred to as the scale effect. This phase is subsequently followed by a decline in emissions as advancements in cleaner technologies and the implementation of regulatory measures come into play, known as the technique effect.

−

IND: This variable quantifies the impact of the industrial sector, encompassing both manufacturing and construction, on the GDP. Given that industrial activities generally require significant energy consumption, this variable functions as an indicator of the underlying structural factors influencing CO₂ emissions within the economic framework. The relationship between GDP and industrial value added aligns with the principles of the EKC, wherein emissions initially increase in tandem with economic growth, subsequently declining as a result of the implementation of cleaner technologies and environmental regulations.

−

EC: It is a consumption originating from fossil fuels, specifically coal, oil, and natural gas. It serves as a direct indicator of the carbon intensity associated with the energy system. This study aims to quantify the extent of fossil energy dependence in landlocked nations.

−

TOP: It is posited to have dual effects: scale effects that lead to an increase in emissions, and technique effects that contribute to their reduction, mediated by its impact on production intensity and logistics demand. In landlocked nations, elevated logistics expenses and reliance on terrestrial transportation could intensify the emissions consequences of trade, unless efficiency improvements are achieved.

−

RE: It is defined by the proportion of total final energy consumption derived from renewable sources, including wind, solar, hydroelectricity, and biomass, as per the classifications established by the IEA (2024) [65]. RE contributes to a reduction in emissions through the decrease in carbon intensity associated with transportation and industrial activities. Nevertheless, the beneficial impact may be attenuated in landlocked nations as a result of infrastructural constraints and limited integration into energy networks.

−

FD: It is measured by domestic credit to the private sector as a percentage of GDP. It is included to assess its role in enabling (or exacerbating) CO₂ emissions. While expanded credit can stimulate investment in cleaner infrastructure, it may also lead to higher emissions through increased consumption and industrial activity, depending on the regulatory environment.

The mechanisms under consideration hold particular significance for limited liability companies, which encounter elevated transportation expenses, reliance on energy imports, and infrastructural constraints.

To ensure that the data are normally distributed, the parameters were converted into logarithmic forms. Details about the variables used in this study, along with the measurement and sources of data, are given in

Table 1. Description of the variables and data sources.

Variables	Measurements	Sources
TCO2	% of total fuel combustion	IEA (2024)
ROFT	Millions of metric tons times kilometers traveled	WDI (2024)
RAFT	Millions of metric tons times kilometers traveled Constant US$	WDI (2024)
GDP		WDI (2024)
IND	Constant US$	WDI (2024)
EC	% of total energy consumption	WDI (2024)
TOP	Million US dollars	WDI (2024)
RE	% of total final energy consumption	WDI (2024)
FD	Domestic credit to private sector (% of GDP)	WDI (2024)

Source: Authors’ Elaboration.

.

3.2. Empirical Model

Using Transport-based CO₂ emissions (TCO₂) as our dependent variable, modeled as a function of ROFT, RAFT, GDP, IND, EC, TOP, RE, and FD, our empirical model is presented as follows:

$${\left({TCO}_2\right)}_{i,t}= f({ROFT}_{i,t};{ RAFT}_{i,t};{ GDP}_{i,t};{ IND}_{i,t};{ EC}_{i,t};{ TOP}_{i,t};{ RE}_{i,t};{ FD}_{i,t})$$

(1)

The econometric model specification is provided as follows in Equation (2):

$${ Ln \left({TCO}_2\right)}_{i,t}= {\alpha }_0+ {\alpha }_1 Ln{ ROFT}_{i,t}+ {\alpha }_2 {Ln RAFT}_{i,t}+ {\alpha }_3 {Ln GDP}_{i,t}+ {\alpha }_4 Ln {IND}_{i,t}+ {\alpha }_5 {Ln EC}_{i,t}+ {\alpha }_6 {Ln TOP}_{i,t}+{ \alpha }_7 {Ln RE}_{i,t}+ {\alpha }_8 {Ln FD}_{i,t}+ {\epsilon }_{i,t}$$

(2)

where ${\alpha }_0 \mathrm{represents} \mathrm{the} \mathrm{constant}, {\alpha }_1$, $\dots$, ${\alpha }_8$ represent the elasticities of the variables in the model with respect to TCO₂. Elasticities are of importance as they bring to bear the actual responses of TCO₂ as the dependent variable against the independent variables. $t$ is the period from 1990 to 2022. $i=1,\dots , 10$ indicates the country index and $\epsilon$, is the classical error term. The log form safeguards the removal of likely outliers as well as large coefficients.

3.3. The Flowchart of the Analysis

This study implements a comprehensive panel econometric methodology to examine the long- and short-run dynamics between transport-related CO₂ emissions and a set of macroeconomic and energy-related variables. First, CD is assessed using the Breusch-Pagan LM test, Pesaran’s scaled LM, bias-corrected scaled LM, and the CD test, which evaluate the interdependence of panel units by analyzing residual correlations. To address parameter heterogeneity, Pesaran and Yamagata [66] slope homogeneity test, comprising the delta ($\tilde{\Delta }$) and adjusted delta ($\tilde{\Delta }$_adj) statistics, is employed to determine whether cointegration slopes are consistent across units. Second-generation panel unit root tests, namely the CADF and the CIPS tests, are used to account for cross-sectional dependencies while testing the stationarity of individual series. Panel cointegration is examined using the Pedroni and Kao residual-based tests. Pedroni’s test allows for heterogeneity in both intercepts and slopes, using seven statistics across within- and between-dimension estimators, while the Kao test assumes homogeneity in cointegration relationships but includes individual fixed effects. The long- and short-run relationships are further estimated using the PMG-ARDL model, which captures the dynamic adjustment of TCO₂ emissions to its equilibrium path. For robustness, FMOLS and DOLS estimators are employed to control for endogeneity and serial correlation in the long-run estimations. Finally, causality is explored using the DH panel causality test, which is suitable for heterogeneous panels with cross-sectional dependence. This approach leverages the average (W⁻) and standardized (Z⁻) statistics to test for Granger-type causality between variables, guiding the formulation of targeted environmental and transport policies based on the directionality of influence among variables. Additionally, the IRF is utilized to trace the dynamic response of TCO₂ emissions to one-time shocks in explanatory variables, offering insights into the magnitude and persistence of the effects over time. Complementarily, VDA quantifies the relative contribution of each explanatory variable to the forecast error variance of TCO₂ emissions, thereby identifying the dominant sources of variation and policy leverage points. This integrated methodological approach ensures robust inference for guiding environmental and transport-related policy formulation in heterogeneous panel settings.

Figure 1 presents the flowchart methodology followed in the empirical examination.

3.4. Cross-Sectional Dependence Test

The empirical study begins with the CD test of the full sample and subsample (Europe and Asia) by employing the Breusch and Pagan LM test, the Pesaran scaled LM test, the Bias-corrected scaled LM test and the Pesaran CD test. Pesaran’s CD test is predicated on the mean pairwise correlation of residuals and is presented as follows:

$$CD= \sqrt{{\displaystyle \frac{2T}{N(N-1)}}} ({\sum }_{i=1}^{N-1}{\sum }_{j=i+1}^N{\widehat{\rho }}_{i,j}) ⟹ N (\mathrm{0,1})$$

(3)

$$\left\{\begin{array}{l}{H}_0: {\widehat{\rho }}_{i,j}=0 {\forall }_{i }\ne j\\ H_1: {\widehat{\rho }}_{i,j}\ne 0 for at least one pair (i,j)\end{array}\right.$$

where ${\widehat{\rho }}_{ij}$ is the pairwise correlation coefficient of residuals between units i and j, N represents the number of cross-sectional units, and T represents the number of time periods. The null hypothesis should be accepted when there is CD in the panel data.

3.5. Slope Homogeneity Test

A slope homogeneity test for panel data models, attributed to Swamy [67], developed the framework to find if slope coefficients of the cointegration equation are homogeneous. Pesaran and Yamagata [66] improved Swamy’s slope homogeneity test, forming two ‘delta’ test statistics: $\tilde{\Delta }$ and $\tilde{\Delta }$_adj.

$$\tilde{\Delta }=\sqrt{N}\left({\displaystyle \frac{N^{-1}\overline{S}-k}{\sqrt{2k}}}\right)-{x^2}_k$$

(4)

$$\tilde{\Delta }\_\mathrm{a}\mathrm{d}\mathrm{j}=\sqrt{N}\left({\displaystyle \frac{N^{-1}\overline{S}-k}{v(T,k)}}\right)~N(\mathrm{0,1})$$

(5)

The function v(T, k) serves as a standardization term, specifically representing the standard deviation, which addresses the small-sample bias inherent in the slope homogeneity test. The test statistic $\tilde{\Delta }$_adj is modified to conform to the standard normal distribution N(0, 1), when the null hypothesis is assumed to be true. Where $N$ is the number of cross-section units, $\mathrm{S}$ denotes the Swamy test statistic, $k$ represents the independent variables, $\tilde{\Delta }$ and $\tilde{\Delta }$_adj are suitable for large and small samples, respectively. The null and alternative hypotheses used for the slope homogeneity test are as follows:

$$\left\{\begin{array}{c}{H}_0: \mathrm{Cointegration} \mathrm{coefficients} \mathrm{are} \mathrm{homogeneous}.\\ H_1: \mathrm{Cointegration} \mathrm{coefficients} \mathrm{are} \mathrm{heterogeneous}.\end{array}\right.$$

3.6. Panel Unit Root Test

In this study, two second-generation panel root tests, specifically the CADF and the CIPS Pesaran tests [68], were employed. The CADF model can be presented as follows in Equation (6):

$$\Delta y_{i,t}= {\alpha }_i+ {\beta }_i{y}_{i,t - 1} + {\gamma }_i{\overline{ y}}_{t-1} + {\sum }_{j=0}^p{\theta }_{i,j }\Delta {\overline{ y}}_{t-j} + {\sum }_{j=1}^p{\delta }_{i,j}\Delta y_{i,t - j}+ {ϵ}_{i,t}$$

(6)

where $\overline{y}$ = $N^{-1}\sum _{j=1}^N{y}_{i,t}$.

The CIPS statistic is the mean value of t-statistics that are obtained from estimating the CADF regression for each cross-section:

$$\mathrm{CIPS} = {\displaystyle \frac{1}{N}}{\sum }_{i=1}^N{CADF}_i$$

(7)

where the CADF_i is the cross-sectional augmented Dickey–Fuller statistics for i^th cross-section unit, and N represents a cross-sectional dimension. In the context of the unit root test, the null hypothesis postulates that all series within the panel follow a non-stationary process, whereas the alternative hypothesis proposes that a fraction of the series in the panel is characterized by stationarity.

3.7. Panel Cointegration Test

This study adopts two panel cointegration tests, namely the Pedroni residual cointegration test [69] and the Kao test [70]. Pedroni [69] introduces statistics that examine the hypothesis of no cointegration in nonstationary panels. These statistics account for both within-dimension (panel statistics) and between-dimension (group statistics) variations, allowing for heterogeneity in the short-run dynamics, as well as in the long-run intercepts and slope coefficients across cross-sectional units. This flexibility makes the Pedroni test particularly suitable for heterogeneous panel structures commonly encountered in macroeconomic and environmental studies.

$${\overline{y}}_t={\displaystyle \frac{1}{N}}{\sum }_{i=1}^N{y}_{i,t}$$

(8)

The entire test statistics are residual-based tests, with residuals collected from the following regressions:

$$y_{i,t}={\alpha }_i+{\beta }_{1i} x_{1i,t}+{\beta }_{2i} x_{2i,t}+\dots +{\beta }_{Mi} x_{Mi,t}+e_{i,t}$$

(9)

$$\Delta y_{i,t}={\sum }_{m=1}^M{\beta }_{mi} \Delta x_{mi,t}+{\eta }_{i,t}$$

(10)

$${\widehat{e}}_{i;t}={\widehat{\gamma }}_i{\widehat{e}}_{i,t-1}+{\widehat{\mu }}_{i,t}$$

(11)

$${\widehat{e}}_{i;t}={\widehat{\gamma }}_i{\widehat{e}}_{i,t-1}+{\sum }_{k=1}^k{\widehat{\gamma }}_{i,k} \Delta {\widehat{e}}_{i,t-k}+{{\widehat{\mu }}^{*}}_{i,t}$$

(12)

where $i=1, 2\dots ,N$ is the number of individuals in the panel, $t=1, 2,\dots ,T$ is the number of time periods, $m=1, 2,\dots ,M$ is the number of regressors, and $k=1, 2,\dots ,K$ represents the quantity of lags of the first-differenced residuals incorporated in the ADF regression, serving the purpose of mitigating autocorrelation.

Kao [70] proposes estimating the homogeneous cointegrating relationship through pooled regression, while accounting for individual fixed effects. The regression equation is as follows:

$$Y_{i,t}={\alpha }_i+\beta X_{i,t}+{\mu }_{i,t}$$

(13)

The least squared dummy variable (LSDV) estimator for $\beta$ is:

$$\tilde{\beta }=({\sum }_{i=1}^N{\sum }_{t=1}^T{\tilde{Y}}_{i,t}{\widetilde{X^{\prime }}}_{i,t}\left)\right({{\sum }_{i=1}^N{\sum }_{t=1}^T{\tilde{X}}_{i,t}{\widetilde{X^{\prime }}}_{i,t})}^{-1}$$

(14)

where ${\widehat{Y}}_{i,t}=Y_{i,t}-{\displaystyle \frac{1}{T}}\sum _{s=1}^T{Y}_{i,s}$ and ${\widehat{X}}_{i,t}=X_{i,t}-{\displaystyle \frac{1}{T}}\sum _{s=1}^T{X}_{i,s}$. The residuals from this first-stage regression ${\tilde{u}}_{i,t}={\tilde{Y}}_{i,t}-\tilde{\beta }{\tilde{X}}_{i,t}$ will still contain a unit root under the null hypothesis of no cointegration.

3.8. PMG-ARDL Model

The specification of panel linear ARDL model (p, q) developed by Pesaran et al. (1999) [71] is given in Equation (15):

$$\mathrm{Ln} {{(TCO}_2)}_{i,t}= {\sum }_{j=1}^p{\lambda }_{i,j} Ln {{(TCO}_2)}_{i,t-j}+ {\sum }_{j=1}^q{\delta }_{i,j }{X}_{i,t-j}+ {\mu }_i+ {\epsilon }_{i,t}$$

(15)

where TCO₂ is the dependent variable and indicates the transport carbon dioxide emission, $X_{i,t}$ is (k × 1) vector of explanatory variables including road freight transport, rail freight transport, gross domestic product, industry, fossil fuel energy consumption, trade openness and renewable energy consumption, ${\mu }_i$ represents the fixed effects, ${\lambda }_{i,j}$ is the coefficient of the lagged dependent variable, ${\delta }_{i,j}$ is (k × 1) coefficient vector of independent variables, ${\epsilon }_{i,t}$ is the error term, i(1, 2, …, N) is the number of cross-section, and t(1, 2,…, T) denotes the time. The absence of a long-run relationship among the variables is represented as null hypothesis. By transforming Equation (15), we obtain the panel ARDL (p, q) error correction model:

$$\Delta Ln{{(TCO}_2)}_{\mathrm{i},\mathrm{t}}= {\mathrm{\varnothing }}_{\mathrm{i}} Ln{\left({TCO}_2\right)}_{\mathrm{i},\mathrm{t}}+{\sum }_{\mathrm{j}=1}^{\mathrm{p}-1}{\lambda }_{i,j} \Delta Ln ({{TCO}_2)}_{\mathrm{i},\mathrm{t}-\mathrm{j}} + {\sum }_{\mathrm{j}=1}^{\mathrm{q}-1}{\delta }_{i,j }{\Delta X}_{i,t-j} + {\mathsf{\mu }}_{\mathrm{i}} + {\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(16)

where $Ln {{(TCO}_2)}_{\mathrm{i},\mathrm{t}}= {\varnothing }_i Ln{ {(TCO}_2)}_{i,t-1} - {\acute{\beta }}_i{X}_{i,t}$, ${\lambda }_{i,j}$ and ${\mathrm{\varnothing }}_{\mathrm{i}}$ provide both the short-run dynamics and the long-run information contained in the data, respectively. Additionally, ${\mathrm{\varnothing }}_{\mathrm{i}}$ denotes the error correction term to evaluate the speed of adjustment of TCO₂ emissions toward its long-run equilibrium with the variation in explanatory variables.

3.9. Robustness Check

The FMOLS showed the ability to tackle endogeneity and serial correlation in long-run relationships. Following Pedroni [72], the panel FMOLS estimator is expressed as:

$${{\widehat{\beta }}^{*}}_{GFM}=N^{-1}{\sum }_{i=1}^N{{\widehat{\beta }}^{*}}_{FM,i}$$

(17)

where the FMOLS estimator is applied to each country and the associated t-statistic is:

$$t_{{\widehat{\beta }}_{GFM}}=N^{{\displaystyle \raisebox{1ex}{$-1$}\!\left/ \!\raisebox{-1ex}{$2$}\right.}}{\sum }_{i=1}^N{t}_{{{\widehat{\beta }}^{*}}_{FM,i}}$$

(18)

The DOLS is a parametric model recommended by Kao [70] and Mark and Sul [73] for the first time. DOLS is appropriate for small samples compared to OLS and FMOLS, and is used for the long-run analysis to take into account the potential heterogeneity and serial correlation. Following Alvarez et al. [74], the DOLS method is expressed as:

$$y_{i,t}=\alpha +x_{i,t} \beta +{\sum }_{j={-q}_1}^{q_2}{c}_{i,j} \Delta x_{i,t+j}+v_{i,t}$$

(19)

The incorporation of leads and lags of $\Delta x_{i,t+j} \mathrm{aims} \mathrm{to}$ asymptotically eradicate potential endogeneity and serial correlation, thus ensuring that the OLS estimator of β remains consistent and asymptotically efficient.

3.10. Dumitrescu Hurlin Panel Causality Test

The DH test is based on an average of individual Wald statistics from N cross-sectional regressions:

$$\overline{W}={\displaystyle \frac{1}{N}}{\sum }_{i=1}^N{W}_i$$

(20)

where $W_i$ is the Wald statistic for unit i.

The standardized statistic is given by:

$$\overline{Z}=\sqrt{N} \left({\displaystyle \frac{\overline{W}-k}{\sqrt{2k}}}\right)$$

(21)

After analyzing the long and short-run relationships, it is crucial to analyze the causal relationship to guide better policies. To ascertain the association between the dependent variable represented by carbon dioxide emissions from transportation and the independent variables in LLCs, we employ the panel causality test developed by Dumitrescu and Hurlin [75]. The advantage of the DH panel causality test is its ability to analyze data in the presence of sectional dependence between countries. This test uses two different statistics: the Wbar statistics and the Zbar statistics. The Wbar statistics take an average of test statistics, while the Zbar statistics designate a standard normal distribution. If the p-value is less than 5%, 10% and 1%, the causality test rejects the null hypothesis.

4. Results and Discussion

4.1. Descriptive Analysis

Table 2. Descriptive statistics.

	Ln TCO2	Ln ROFT	Ln RAFT	Ln FD	Ln GDP	Ln IND	Ln EC	Ln TOP	Ln RE
All sample LLCs
Mean	2.869	9.495	9.436	3.677	8.914	21.652	5.040	11.302	1.835
Maximum	4.224	12.115	12.648	5.194	11.803	25.988	11.833	14.105	3.951
Minimum	1.537	4.889	5.948	0.009	4.098	2.641	3.840	8.315	−0.328
Std. Dev.	0.729	1.393	1.407	1.021	1.590	6.333	1.950	1.392	1.185
European LLCs
Mean	3.158	9.651	9.227	4.117	9.599	21.018	5.233	2.433	11.929
Maximum	4.232	11.460	11.230	5.227	11.803	26.122	11.833	3.951	14.105
Minimum	1.548	5.624	5.948	1.360	6.387	2.641	3.808	−0.083	8.315
Std. Dev.	0.624	1.087	0.909	0.749	1.359	7.472	2.310	0.931	1.216
Asian LLCs
Mean	2.333	8.814	9.375	2.778	7.499	23.005	4.512	10.063	0.938
Maximum	3.626	12.115	12.648	4.369	9.538	25.070	4.612	11.960	2.876
Minimum	0.641	4.889	5.777	0.009	4.098	20.448	4.054	8.344	−0.328
Std. Dev.	0.668	2.105	2.204	0.893	1.120	1.216	0.142	0.978	0.772

Source: Authors’ own calculation by using E-Views 12.

provides a comparative analysis of descriptive data for all sample, European and Asian LLCs. A general overview of the descriptive statistics reveals notable regional differences across the variables. On average, European LLCs report higher total carbon emissions (Ln TCO₂ = 3.158) and GDP levels (Ln GDP = 9.599) than their Asian counterparts (Ln TCO₂ = 2.333; Ln GDP = 7.499), reflecting greater economic maturity and associated environmental externalities. Renewable energy use in European LLCs is significantly higher (mean = 11.929) compared to Asian LLCs (mean = 0.938), indicating a stronger commitment to green energy transition. Standard deviation values reveal substantial variability in industrial activity, particularly in European LLCs (Std. Dev. = 7.472) and the full sample (Std. Dev. = 6.333), suggesting a wide range of industrialization levels across countries.

4.2. Cross-Sectional Dependence Test Results and Discussion

As shown in

Table 3. Results of cross-sectional dependence analysis test for all sample, European and Asian LLCs.

	Breusch-Pagan LM	Pesaran Scaled LM	Bias-Corrected Scaled LM	Pesaran CD
All sample LLCs
Ln TCO2	776.236 (0.000) ***	68.767 (0.000) ***	68.595 (0.000) ***	21.246 (0.000) ***
Ln ROFT	735.059 (0.000) ***	64.841 (0.000) ***	64.669 (0.000) ***	24.921 (0.000) ***
Ln RAFT	435.153 (0.000) ***	36.246 (0.000) ***	36.074 (0.000) ***	6.876 (0.000) ***
Ln FD	370.518 (0.000) ***	30.083 (0.000) ***	29.911 (0.000) ***	11.440 (0.000) ***
Ln GDP	1542.792 (0.000) ***	141.855 (0.000) ***	141.683 (0.000) ***	39.187 (0.000) ***
Ln IND	1085.402 (0.000) ***	98.244 (0.000) ***	98.073 (0.000) ***	31.677 (0.000) ***
Ln EC	650.825 (0.000) ***	56.809 (0.000) ***	56.637 (0.000) ***	10.302 (0.000) ***
Ln TOP	917.698 (0.000) ***	82.255 (0.000) ***	82.083 (0.000) ***	28.152 (0.000) ***
Ln RE	461.787 (0.000) ***	38.785 (0.000) ***	38.613 (0.000) ***	15.105 (0.000) ***
European LLCs
Ln TCO2	476.746 (0.000) ***	70.323 (0.000) ***	70.213 (0.000) ***	21.753 (0.000) ***
Ln ROFT	269.915 (0.000) ***	38.408 (0.000) ***	38.299 (0.000) ***	14.507 (0.000) ***
Ln RAFT	197.594 (0.000) ***	27.249 (0.000) ***	27.139 (0.000) ***	2.525 (0.011) **
Ln FD	102.947 (0.000) ***	12.644 (0.000) ***	12.535 (0.000) ***	4.072 (0.000) ***
Ln GDP	629.852 (0.000) ***	93.947 (0.000) ***	93.838 (0.000) ***	25.081 (0.000) ***
Ln IND	387.646 (0.000) ***	56.574 (0.000) ***	56.465 (0.000) ***	18.605 (0.000) ***
Ln EC	293.971 (0.000) ***	42.120 (0.000) ***	42.011 (0.000) ***	15.467 (0.000) ***
Ln TOP	343.282 (0.000) ***	49.729 (0.000) ***	49.619 (0.000) ***	16.981 (0.000) ***
Ln RE	412.020 (0.000) ***	60.335 (0.000) ***	60.226 (0.000) ***	19.988 (0.000) ***
Asian LLCs
Ln TCO2	34.127 (0.000) ***	8.119 (0.000) ***	8.057 (0.000) ***	0.645 (0.518)
Ln ROFT	69.731 (0.000) ***	18.397 (0.000) ***	18.335 (0.000) ***	7.984 (0.000) ***
Ln RAFT	26.741 (0.000) ***	5.987 (0.000) ***	5.925 (0.000) ***	1.517 (0.129)
Ln FD	49.639 (0.000) ***	12.597 (0.000) ***	12.535 (0.000) ***	6.676 (0.000) ***
Ln GDP	160.542 (0.000) ***	44.612 (0.000) ***	44.550 (0.000) ***	12.643 (0.000) ***
Ln IND	147.389 (0.000) ***	40.815 (0.000) ***	40.753 (0.000) ***	12.117 (0.000) ***
Ln EC	49.186 (0.000) ***	12.466 (0.000) ***	12.404 (0.000) ***	−1.154 (0.248)
Ln TOP	116.111 (0.000) ***	31.786 (0.000) ***	31.723 (0.000) ***	10.600 (0.000) ***
Ln RE	24.201 (0.000) ***	5.254 (0.000) ***	5.191 (0.000) ***	−1.097 (0.272)

Source: Authors’ own calculation by using E-Views 12. Notes: Null hypothesis: No cross-section dependence (correlation). *** and ** indicate rejection of the null hypothesis at 1% and 5% levels of significance, respectively. Numbers in parentheses are p-values.

, all statistical tests are significant at the 1% level, which leads to the rejection of the null hypothesis of no cross-sectional dependence and acceptance of the alternative hypothesis of cross-sectional dependence at the 1% significance level in the panel variables. This study examines CD to assess the transmission of economic or environmental shocks from one country to others within the sample. This step guarantees the selection of appropriate panel data tests that consider inter-country relationships. These findings imply that the variables of LLCs are cross-sectionally dependent, suggesting that shocks occurring in one of the 10 LLCs seem to be transmitted to other countries. In order to tackle this significant matter, the null hypothesis of the test posits that CD is not present in the data.

4.3. Slope Homogeneity Test Results and Discussion

Table 4. Slope homogeneity test for all sample, European and Asian LLCs.

	All sample LLCs		European LLCs		Asian LLCs
Statistics	T-Statistics	p-Value	T-Statistics	p-Value	T-Statistics	p-Value
$\tilde{\Delta }$	12.255	0.0001 ***	17.698	0.0000 ***	12.084	0.0000 ***
$\tilde{\Delta }$_adj	18.053	0.000 ***	12.292	0.0005 ***	16.115	0.0000 ***

Source: Authors’ own calculation by using E-Views 12. Notes: The Null hypothesis for the slope homogeneity test is that slope coefficients are homogeneous. *** denotes significant at 1% level.

reports the results of Pesaran and Yamagata’s [66] slope homogeneity test. It is shown that the null hypothesis is significant at 1% level, confirming the homogeneous data. The delta test results confirm that slope coefficients are heterogeneous in the CD, and the model of the analysis has a heterogeneity problem. This study examines slope homogeneity to ascertain whether the long-run relationships exhibit uniformity across all nations or demonstrate variation contingent upon individual countries. The significance of this matter lies in the considerable differences in economic structures and transport infrastructures among LLCs. The results of these tests confirm the CD and slope heterogeneity in the sample of LLCs. We consider the second-generation panel unit-root tests more appropriate for our analysis. They should provide a concise and precise description of the experimental results, their interpretation, and the experimental conclusions that can be drawn.

4.4. Panel Unit Root Test Results and Discussion

Table 5. Second-generation panel unit root test.

	CIPS		CADF
	Level	1st Difference	Level	1st Difference
All sample
Ln TCO2	−2.273 **	−5.774 ***	−1.564	−4.431 ***
Ln GDP	−3.476 ***	−5.254 ***	−2.557 ***	−3.986 ***
Ln RAFT	−1.908	−5.317 ***	−1.935	−4.101 ***
Ln ROFT	−1.816	−5.337 ***	−1.454	−4.059 ***
Ln FD	−1.788	−5.175 ***	−1.992	−3.752 ***
Ln IND	−1.798	−4.312 ***	−2.402 **	−3.647 ***
Ln TOP	−1.794	−5.167 ***	−1.687	−4.230 ***
Ln EC	−1.371	−5.276 ***	−1.201	−3.767 ***
Ln RE	−1.973	−4.989 ***	−2.110	−3.326 ***
Europe
Ln TCO2	−3.269 ***	−5.776 ***	−2.469 **	−4.315 ***
Ln GDP	−4.193 ***	−5.933 ***	−1.912	−4.404 ***
Ln RAFT	−1.747	−5.018 ***	−1.742	−3.717 ***
Ln ROFT	−1.934	−5.946 ***	−2.228	−4.133 ***
Ln FD	−1.640	−5.053 ***	−1.630	−3.806 ***
Ln IND	−0.812	−5.151 ***	−1.067	−4.322 ***
Ln TOP	−1.515	−5.284 ***	−1.651	−3.478 ***
Ln EC	−2.077	−5.365 ***	−1.822	−4.130 ***
Ln RE	−2.032	−5.320 ***	−1.536	−4.119 ***
Asia
Ln TCO2	−2.856 ***	−5.442 ***	−2.131	−4.589 ***
Ln GDP	−2.062	−4.970 ***	−2.229	−4.223 ***
Ln RAFT	−1.832	−5.696 ***	−2.839 **	−4.662 ***
Ln ROFT	−1.913	−5.539 ***	−1.865	−4.589 ***
Ln FD	−1.946	−5.485 ***	−2.766 **	−4.172 ***
Ln IND	−1.742	−4.845 ***	−2.391 *	−3.576 ***
Ln TOP	−1.896	−5.454 ***	−2.087	−4.252 ***
Ln EC	−2.237 *	−6.072 ***	−1.690	−4.757 ***
Ln RE	−2.551 **	−5.272 ***	−2.562 **	−4.280 ***

Source: Authors’ own calculation by using E-Views 12. Notes: Null hypothesis indicates that time series are not stationary. ***, ** and * represent 1%, 5% and 10% significance levels, respectively.

displays the results of the unit root test in both the level and first difference. The results of the second-generation panel unit root tests (CIPS and CADF) reveal a mixed order of integration across the variables. In light of the presence of CD and slope heterogeneity, second-generation panel unit root tests are employed to ascertain the stationarity characteristics of the variables, thereby mitigating potential bias arising from cross-country correlations. According to the CIPS test, in the full sample, all variables are non-stationary at the same level except for GDP and TCO₂, a pattern mirrored in the European subsample. In Asia, however, EC, RE, and TCO₂ are stationary at the same level. The CADF test shows that for the full sample, all variables are non-stationary at the same level except for GDP and IND, while in Europe, only TCO₂ is stationary. In Asia, RAFT, FD, and RE are stationary at the same level. These findings indicate that some variables are I(0), while others are I(1), confirming a combination of different integration levels. The consistency between the CIPS and CADF tests supports the suitability of the PMG-ARDL approach, which can handle variables with mixed integration orders, for robust and reliable analysis.

4.5. Panel Cointegration Tests Results and Discussion

Table 6. Results of the cointegration tests for all sample, European and Asian LLCs.

	All Sample LLCs		European LLCs		Asian LLCs
Pedroni Residual Cointegration Test
Alternative hypothesis: common AR coefs. (within-dimension)
	Statistic	p-value	Statistic	p-value	Statistic	p-value
Panel v-Statistic	−2.244	0.987	−0.300	0.618	−1.857	0.968
Panel rho-Statistic	2.480	0.993	1.182	0.881	1.826	0.966
Panel PP-Statistic	−0.751	0.226	−2.482	0.006 ***	0.321	0.626
Panel ADF-Statistic	−1.004	0.157	−3.543	0.000 ***	0.537	0.704
Alternative hypothesis: individual AR coefs. (between-dimension)
Panel rho-Statistic	2.983	0.998	2.054	0.980	2.229	0.987
Panel PP-Statistic	−2.265	0.011 **	−2.165	0.015 **	−0.892	0.186
Panel ADF-Statistic	−0.033	0.486	−0.785	0.216	0.983	0.837
Kao Residual Cointegration Test
	t-Statistic	Prob.	t-Statistic	Prob.	t-Statistic	Prob.
ADF	−1.267	0.102	−3.190	0.000 ***	−4.745	0.000 ***

Source: Authors’ own calculation by using E-Views 12. Note: *** and ** denote statistical significance levels at 1% and 5%, respectively.

reports the panel cointegration tests results. The Pedroni cointegration test results reveal that the panel PP-Statistic and the panel ADF-Statistic exhibit significant findings at a 1% level of significance. The analysis of long-run cointegration, utilizing the Pedroni and Kao tests, serves to ascertain whether the variables exhibit a tendency to move in tandem over time, thereby suggesting the presence of stable long-run relationships. As a result, we can conclude that there is evidence of cointegration between the variables. Furthermore, the ADF statistics of the Kao test are found to be significant at a 1% significance level. Thus, the study confirms the stable relationship between the variables in the long run and eliminates the possibility of spurious regression in the model. The findings of this study suggest that any disruption to sustainable transportation will have a lasting impact on carbon dioxide emissions. Therefore, it is crucial to assess the long-term and short-term impacts by utilizing the PMG estimator within the framework of the ARDL model. This section may be divided by subheadings. It should provide a concise and precise description of the experimental results, their interpretation, and the experimental conclusions that can be drawn.

4.6. Results of the PMG-ARDL Estimation and Discussion

As reported in

Table 7. The results of PMG-ARDL for all sample, European and Asian LLCs.

	All Sample LLCs		European LLCs		Asian LLCs
Variables	Coef.	p-Value	Coef.	p-Value	Coef.	p-Value
Long-run analysis
Ln ROFT	0.053	0.568	0.874	0.000 ***	0.251	0.000 ***
Ln RAFT	−0.690	0.000 ***	0.017	0.658	−0.806	0.000 ***
Ln FD	0.650	0.000 ***	−1.204	0.000 ***	0.535	0.000 ***
Ln GDP	0.720	0.000 ***	−0.355	0.000 ***	0.318	0.000 ***
Ln IND	−1.104	0.000 ***	1.327	0.000 ***	−0.525	0.000 ***
Ln EC	−3.362	0.000 ***	0.303	0.318	12.595	0.000 ***
Ln TOP	−0.394	0.001 ***	−0.208	0.000 ***	0.074	0.468
Ln RE	0.176	0.129	0.368	0.000 ***	0.512	0.000 ***
Short-run analysis
Coint Eq (-1)	−0.108	(0.053) **	−0.121	0.024 **	−0.410	0.025 **
D(Ln ROFT)	0.012	(0.889)	−0.168	0.396	−0.042	0.631
D(Ln RAFT)	0.032	(0.574)	−0.049	0.504	0.147	0.590
D(Ln FD)	−0.213	(0.293)	0.594	0.443	0.298	0.469
D(Ln GDP)	0.037	(0.368)	0.030	0.661	−0.276	0.014 **
D(Ln IND)	0.062	(0.635)	−0.062	0.712	−0.125	0.675
D(Ln EC)	−0.303	(0.579)	0.071	0.908	−6.755	0.273
D(Ln TOP)	0.078	(0.012) **	−0.180	0.417	−0.016	0.896
D(Ln RE)	0.054	(0.423)	0.166	0.268	−0.037	0.663
C	4.871	(0.057) **	−3.352	0.265	−17.395	0.025 *

Source: Authors’ own calculation by using E-Views 12. Notes: ***, **, and * denote statistically significant levels at 1%, 5% and 10%, respectively. The model selection method used is the Akaike information criterion: Full sample selected model: ARDL(2, 1, 1, 1, 1, 1, 1, 1, 1); Europe selected model: ARDL(1, 2, 2, 2, 2, 2, 2, 2, 2); Asia selected model: ARDL(2, 2, 2, 2, 2, 2, 2, 2, 2).

, the results of the PMG-ARDL estimation provide valuable insights into the determinants of transport-related CO₂ emissions (TCO₂) across the full sample, Europe, and Asia, highlighting both long-run and short-run dynamics. In order to assess the short- and long-term effects, the PMG-ARDL model is employed. This model is capable of handling a combination of stationary and non-stationary variables while also permitting the incorporation of country-specific short-term dynamics. For the full sample, FD and GDP exhibit a positive and significant long-run relationship with TCO₂, suggesting that economic growth and financial expansion contribute to increased emissions. Conversely, RAFT, IND, EC, and TOP demonstrate a negative and significant impact on TCO2, indicating their potential role in mitigating emissions. Notably, ROFT and RE show a positive but insignificant relationship with TCO₂, diverging from prior studies such as You et al. [76]. In the short run, the error correction term is negative and significant, confirming the model’s convergence to equilibrium, while TOP is the only variable with a significant and positive impact on TCO₂. For Europe, the long-run results reveal that ROFT, IND, and RE significantly increase TCO₂, while FD, GDP, and TOP have a significant and negative effect, underscoring the region’s unique economic and industrial dynamics. In Asia, all variables except TOP are significant in the long run, with ROFT, FD, GDP, RE, and EC increasing TCO₂, while RAFT and IND reduce emissions. This suggests that rail freight and industrial activities in Asia play a crucial role in lowering emissions, potentially due to efficiency gains or cleaner technologies. The short-run analysis for Asia highlights GDP as the only significant variable, with a negative impact on TCO₂, reflecting the region’s transitional economic adjustments. Across all samples, the negative and significant error correction terms indicate that the TCO₂ models converge to their long-run equilibrium, reinforcing the robustness of the findings. These results align with prior literature, such as Hasanov et al. [77], while also offering region-specific insights. Policymakers should consider these findings to design targeted strategies, such as promoting rail freight transport and renewable energy in Asia, enhancing industrial efficiency in Europe, and balancing economic growth with environmental sustainability globally. Figure 2 further illustrates the regional variations in the long-run determinants of TCO₂, emphasizing the need for tailored policy interventions to address the diverse challenges of transport-related emissions.

4.7. Robustness Check Results and Discussion

Table 8. FMOLS and DOLS estimates for all sample, European and Asian LLCs.

	All Sample LLCs	European LLCs	Asian LLCs
Variables	Coefficient	Coefficient	Coefficient
FMOLS
Ln TCO2	-	-	-
Ln ROFT	−0.019 (0.000) ***	3.95 10−5 (0.997)	0.164 (0.000) ***
Ln RAFT	−0.243 (0.000) ***	0.027 (0.099) *	0.154 (0.005) ***
Ln FD	−0.202 (0.000) ***	−0.248 (0.000) ***	0.171 (0.003) ***
Ln GDP	0.382 (0.000) ***	0.544 (0.000) ***	0.293 (0.000) ***
Ln IND	0.066 (0.000) ***	−0.081 (0.000) ***	−0.305 (0.000) ***
Ln EC	0.119 (0.000) ***	−0.157 (0.000) ***	−0.473 (0.000) ***
Ln TOP	0.030 (0.000) ***	0.080 (0.000) ***	−0.249 (0.000) ***
Ln RE	0.098 (0.000) ***	0.023 (0.000) ***	−0.419 (0.000) ***
DOLS
Ln TCO2	-	-	-
Ln ROFT	−0.665 (0.000) ***	−0.014 (0.764)	0.391 (0.001) ***
Ln RAFT	1.319 (0.000) ***	0.135 (0.006) ***	−0.252 (0.004) ***
Ln FD	2.059 (0.000) ***	−0.326 (0.003) ***	0.255 (0.011) *
Ln GDP	0.848 (0.000) ***	0.527 (0.000) ***	0.518 (0.000) ***
Ln IND	−0.218 (0.594)	−0.086 (0.001) ***	−1.296 (0.000) ***
Ln EC	0.423 (0.032) **	−0.195 (0.003) ***	5.145 (0.000) ***
Ln TOP	0.203(0.112)	0.086 (0.107)	0.306 (0.025) *
Ln RE	−2.070 (0.000) ***	0.048 (0.429)	0.289 (0.351)

Source: Authors’ own calculation by using E-Views 12. Note: ***, **, and * denote statistical significance levels at 1%, 5% and 10%, respectively.

reports the estimation results of FMOLS for full, European and Asian LLCs, which are in line with the baseline estimations. This study presents a comparative analysis of results obtained using FMOLS and DOLS estimators, which are designed to address potential endogeneity and serial correlation in long-run estimates. In Table 8, as one would expect, using each of the three samples, we find that the effect of indicators for sustainable transportation on carbon dioxide emissions from LLCs is larger and statistically significant at the 1% level, except for road freight transport in European LLCs, which has a positive impact but is not significant. Similar to FMOLS estimations, results from DOLS show the influence of the variables employed in the study on carbon dioxide emissions from transport in LLCs. Except for EC and TOP in the entire sample, ROFT in the Europe sample and RE in the Asian sample, all other indicators have a highly significant impact on TCO₂. Overall, our robustness analysis confirms that indicators for sustainable transportation have a strong effect on carbon dioxide emissions from LLCs for both the entire sample and subsamples in the long run. While the effect of the variable under consideration is highly significant in the long run for different estimates (ARDL, FMOLS and DOLS), its effect is weakly significant in the short run. Additionally, in the short run, the entire sample, Europe and Asia respond to different extents. Finally, none of the estimated coefficients is significant for the case of Europe in the short run.

4.8. Results of the Dumitrescu Hurlin Panel Causality Test and Discussion

The results of the DH test are reported in

Table 9. Dumitrescu Hurlin Panel Causality Test.

All Sample LLCs				European LLCs			Asian LLCs
Null Hypothesis	W-Stat.	Zbar-Stat.	Prob.	W-Stat.	Zbar-Stat.	Prob.	W-Stat.	Zbar-Stat.	Prob.
Ln RAFT ≠ Ln TCO2	5.231	2.075	0.037 **	17.468	3.410	0.000 ***	46.386	7.159	8.10−13 ***
Ln TCO2 ≠ Ln RAFT	7.207	4.208	3.10−5 ***	25.300	6.407	1.10−10 ***	26.934	5.316	1.10−7 ***
Ln ROFT ≠ Ln TCO2	4.483	1.273	0.203	11.966	3.749	0.000 ***	24.519	2.776	0.005 ***
Ln TCO2 ≠ Ln ROFT	8.981	6.139	8.10−10 ***	15.924	2.819	0.004 ***	10.175	−0.098	0.921
Ln FD ≠ Ln TCO2	8.405	5.543	3.10−8 ***	3.508	4.006	6.10−5 ***	92.597	16.422	0.000 ***
Ln TCO2 ≠ Ln FD	4.920	1.758	0.0786 *	3.932	4.705	3.10−6 ***	34.068	4.690	3.10−6 ***
Ln GDP ≠ Ln TCO2	6.923	3.934	8.10−5 ***	4.780	2.936	0.003 ***	13.145	0.496	0.619
Ln TCO2 ≠ Ln GDP	4.853	1.686	0.0917 *	4.536	2.662	0.007 ***	9.236	−0.286	0.774
Ln IND ≠ Ln TCO2	9.886	6.939	4.10−12 ***	3.273	3.620	0.000 ***	23.052	10.099	0.000 ***
Ln TCO2 ≠ Ln IND	6.256	3.098	0.001 ***	3.858	4.582	5.10−6 ***	9.759	2.884	0.003 ***
Ln EC ≠ Ln TCO2	2.568	3.062	0.002 ***	1.599	−0.637	0.523	53.231	8.531	0.000 ***
Ln TCO2 ≠ Ln EC	0.955	−0.248	0.803	4.465	2.582	0.009 **	35.371	4.951	710−7 ***
Ln TOP ≠ Ln TCO2	3.847	0.455	0.648	16.776	0.191	0.847	7.075	−0.719	0.471
Ln TCO2 ≠ Ln TOP	2.984	−0.394	0.693	41.008	2.809	0.005 ***	15.945	1.058	0.290
Ln RE ≠ Ln TCO2	2.456	2.828	0.004 ***	24.182	3.584	0.000 ***	34.954	4.868	1.10−6 ***
Ln TCO2 ≠ Ln RE	2.033	1.959	0.050 **	27.921	4.575	5.10−6 ***	15.446	0.958	0.338
Summary of causalities: All sample LLCs: RAFT ↔ TCO2; TCO2 → ROFT; FD ↔ TCO2; GDP ↔ TCO2; IND ↔ TCO2; EC → TCO2; TCO2 ↔ RE European LLCs: ROFT ↔ TCO2; RAFT ↔ TCO2; FD ↔ TCO2; GDP ↔ TCO2; IND ↔ TCO2; TCO2 → EC; TCO2 → TOP; RE ↔ TCO2 Asian  LLCs: RAFT↔TCO2; ROFT→ TCO2; FD ↔ TCO2; IND ↔ TCO2; EC ↔ TCO2; RE → TCO2

Source: Authors’ own calculation by using E-Views 12. Notes: ***, ** and * denote statistical significance levels at 1%, 5% and 10%, respectively. → indicates unidirectional causality, ↔ indicates bidirectional causality and ≠ denotes that A does not homogeneously cause B.

and Figure 2. The DH test, developed by Dumitrescu and Hurlin [75], constitutes a robust extension of the traditional Granger causality test adapted for heterogeneous panel data. This methodology offers the significant advantage of allowing coefficient variation across cross-sectional units (countries) while maintaining the assumption of temporal stationarity [78]. This study investigates causal directions utilizing the Dumitrescu-Hurlin test to ascertain whether alterations in one variable may influence modifications in another.

In the full sample LLCs, bidirectional causality is observed between TCO₂ and RAFT, FD, GDP, IND, and RE, supporting the findings of Rashid Khan et al. [79], who also identified similar bidirectional links. A unique unidirectional causality from TCO₂ to ROFT suggests that changes in emissions influence road freight activity, echoing McKinnon’s [80] conclusions about the regulatory impact of decarbonization efforts. Additionally, EC is found to unidirectionally cause TCO₂, consistent with the IEA’s (2021) [81] observations on fossil fuel dependency in transport. In European LLCs, the causal relationships are more interconnected. TCO₂ has bidirectional causality with all key variables (ROFT, RAFT, FD, GDP, IND, RE), reflecting deeper integration of environmental and economic policies across the EU. Interestingly, TCO₂ also unidirectionally causes EC, indicating that emission variations influence fossil energy use, likely due to carbon reduction policies that drive cleaner technologies [82]. Another unique finding in European LLCs is the unidirectional causality from TCO₂ to TOP, implying that emission concerns may shape trade behaviors through mechanisms like carbon tariffs or consumer demand for low-emission goods.

4.9. Dynamic Impact Analysis

Figure 3, Figure 4 and Figure 5 display the IRF conducted for the full sample, European, and Asian LLCs. Across all samples, a positive response of ROFT to TCO₂ shocks underscores the strong linkage between freight intensity and emissions, although the effect stabilizes over time, especially in Europe, suggesting greater long-term regulation or infrastructure efficiency [83]. This aligns with findings from Saha et al. [84], who underscore the environmental inefficiency of road-centric freight systems in Bangladesh and advocate for modal shifts. Conversely, RAFT reacts differently across regions: while the full sample shows a modest short-term increase, European countries exhibit a positive but muted response, whereas Asian countries demonstrate a negative and declining trend, indicating that rail serves as a more climate-friendly alternative in Asia. GDP initially declines in response to TCO₂ shocks in the full sample, suggesting emission-related economic costs. This effect is more pronounced in Europe, where environmental regulations may act more stringently, while in Asia, the effect is slightly weaker [85]. Industrial output (Ln IND) generally rises with emissions, more strongly in Asia, reflecting a carbon-intensive industrial structure, whereas Europe’s weaker response may be due to higher energy efficiency or cleaner production technologies [86]. Fossil fuel consumption (Ln EC) displays a robust and consistent positive response in all samples, confirming transport’s dependence on fossil energy [87]. However, renewable energy consumption (Ln RE) responds negatively or insignificantly, especially in Asia, suggesting a sluggish green energy transition; only Europe shows a sustained negative response, hinting at stronger renewable integration mechanisms [88]. This is supported by Hariyani et al. [89], who found that trade openness and financial development also influence emissions dynamics in East Asia-Pacific economies. Trade openness (Ln TOP) exhibits a marginal response to TCO₂ shocks, demonstrating a mildly negative correlation in Europe and a slightly positive correlation in Asia. This suggests that emissions may have a more pronounced dampening effect on trade in regions with stricter environmental policies. This observation is consistent with the principles outlined in trade-emissions theory, which posits that in environments characterized by stringent regulations, the costs associated with environmental compliance may restrict specific export activities. Conversely, in contexts with less regulatory oversight, the expansion of trade could lead to an increase in emissions, attributable to carbon-intensive transportation and production methods [90,91]. The initial contraction of financial development (Ln FD) following emissions shocks is observed across all samples, with a pronounced effect in Asia. However, a gradual rebound is noted, indicating that financial systems exhibit an adaptive response over time to environmental externalities [92]. This pattern illustrates the bifurcated function of finance: in the immediate context, increasing emissions may lead to a constriction of credit for carbon-heavy industries, particularly in response to environmental policy demands; conversely, in the long term, financial frameworks may shift capital allocation towards cleaner technologies, renewable energy initiatives, and sustainable transportation infrastructure [93,94,95]. The interplay between these contrasting mechanisms, scale effects that may elevate emissions and efficiency effects that may diminish them, plays a crucial role in ascertaining whether financial development ultimately intensifies or alleviates CO₂ emissions. These findings point to the persistent carbon dependency of freight systems and economic structures in LLCs, while also revealing regional asymmetries in green transition trajectories. For policy, a modal shift from road to rail, enhanced renewable integration, support for green finance, and industrial decarbonization remain central to reconciling logistical efficiency with environmental sustainability.

As reported in

Table 10. Variance decomposition analysis.

Period	S.E.	Ln TCO2	Ln ROFT	Ln RAFT	Ln FD	Ln GDP	Ln IND	Ln EC	Ln TOP	Ln RE
All sample LLCs
2023	0.101	100.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
2028	0.193	96.709	1.040	0.348	0.724	0.174	0.324	0.011	0.173	0.492
2033	0.251	90.176	1.196	3.684	1.324	0.209	0.426	0.047	0.994	1.940
European LLCs
2023	0.081	100.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
2028	0.164	90.589	4.520	1.497	0.089	0.942	1.426	0.853	0.017	0.063
2033	0.205	84.927	7.7889	1.408	0.108	1.238	1.994	2.019	0.436	0.077
Asian LLCs
2023	0.226	100.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000
2028	0.461	74.878	1.930	9.988	2.024	3.634	1.518	0.992	0.047	4.986
2033	0.5493	60.925	1.7838	13.623	6.345	7.957	1.651	3.163	0.250	4.299

Source: Authors’ own calculation by using E-Views 12.

, the VDA of TCO₂ reveals a dynamic temporal evolution and pronounced regional disparities that carry important implications for environmental policy. Across all regions, the contribution of TCO₂’s own shocks to its forecast variance declines significantly from 2023 to 2033, indicating that emissions become increasingly influenced by external factors over time. This trend is most pronounced in Asia, where self-influence falls from 100% to just 60.93%, compared to 90.18% globally and 84.93% in Europe, suggesting a growing complexity in the determinants of emissions. Simultaneously, the rising standard errors underscore heightened forecast uncertainty, further emphasizing the multifactorial nature of CO₂ emissions. Regionally, the drivers of TCO₂ emissions differ markedly: in Europe, ROFT is the dominant external factor, with moderate contributions from EC and IND, while RE plays a marginal role. In contrast, Asia exhibits a broader and more balanced mix of contributors, with RAFT, GDP, FD, RE, and EC all exerting significant influence, reflecting the region’s diverse development paths and energy structures. The global sample presents an intermediate pattern, where RAFT, RE, FD, and TOP moderately explain emission variance. These findings suggest that effective decarbonization policies must be regionally tailored: Europe should focus on decarbonizing road freight and improving industrial energy efficiency, while Asia requires a multifaceted strategy involving rail modernization, green finance, renewable energy expansion, and economic decoupling. Globally, a coordinated policy mix targeting transport, energy, and financial systems is essential to drive emissions reductions. Overall, the analysis underscores the decreasing dominance of historical emission patterns and the growing role of sectoral and economic dynamics, highlighting the necessity for adaptive, region-specific approaches to climate governance.

5. Conclusions and Policy Implications

This study explores the dynamic impact of logistic operations, financial development, fossil fuel energy consumption, renewable energy consumption, and industrialization on transport-based carbon dioxide emissions in LLCs across Europe and Asia. Despite a growing body of literature on the causal relationship between transport emissions and indicators for sustainable transportation in recent years, no previous study has explored this specific interrelationship. This study aims to address this research gap by investigating the causal nexus between transport-related CO₂ emissions, economic growth, road freight transport, rail freight transport, industry, trade openness, fossil fuel energy consumption, financial development, and renewable energy consumption in ten LLCs from 1990 to 2022.

Our study’s originality lies in its disaggregation of freight transport into road and rail modes, revealing stark contrasts in their environmental impacts, and its identification of distinct region-specific dynamics, with rail freight consistently mitigating emissions in Asia but showing limited influence in Europe. We further uncover the complex role of financial development, which is associated with higher emissions in Asia but lower emissions in Europe, suggesting divergent investment pathways. Additionally, we reveal the nuanced role of renewable energy, which proves effective in reducing CO₂ emissions in Europe but not yet in Asia, likely due to infrastructural and transitional inefficiencies. By integrating causality analysis, we confirm bidirectional relationships between CO₂ emissions and key economic variables, underscoring feedback loops often overlooked in earlier LLC studies. The findings of the PMG estimation for all sample LLCs demonstrate that rail freight transport is highly significant and has a negative relationship with TCO₂. This result means that a percentage point increase in rail freight transport will cause TCO₂ to decrease by 0.690% in the long run. Industry, energy consumption, and trade openness variables also reveal a negative relationship with TCO₂ in the long run. This means that a percentage point increase in IND, EC, and TOP will cause TCO₂ to decrease by 1.104%, 3.362%, and 0.394%, respectively. There is a positive but insignificant relationship between road freight transport, renewable energy consumption, and TCO₂ in the long run. In sum, for the all sample LLCs, economic growth and financial development are the main causes of environmental degradation in the transport sector. More interestingly, rail freight transport, industry, trade openness, and fossil fuel energy consumption have a remarkable negative impact on transport carbon emissions. The findings of the PMG estimation for the European sample of LLCs demonstrate that ROFT, IND, and RE significantly increase TCO₂ in the long run. When these variables grow by 1%, transport carbon dioxide emissions increase by 0.874%, 1.327%, and 0.368%, respectively. However, the impact of FD, GDP, and TOP on TCO₂ emissions is significantly negative. The findings of the PMG estimation for the Asia sample of LLCs demonstrate that all the estimated variable coefficients are significant at the 1% level, except for TOP, which has a positive but insignificant impact on TCO₂. Furthermore, the findings reveal that ROFT, FD, GDP, RE, and EC increase TCO₂ in the long term; however, RAFT and IND decrease TCO₂. The error correction terms for all sample countries, Europe, and Asia are found to be negative and significant. Robustness checks using FMOLS and DOLS confirm the long-term significance of the majority of explanatory variables, highlighting the validity of the PMG findings. All models exhibit negative and statistically significant error correction factors, hence affirming the presence of long-term equilibrium relationships.

The findings of this paper carry several important implications for policymakers and strategies in LLCs, which can be implemented to tackle the challenges faced and to improve their logistics connectivity:

• Invest in rail infrastructure: Given that rail freight transport substantially decreases total CO₂ emissions over time, landlocked nations should prioritize the expansion and modernization of rail networks to transition freight from road-based transport systems.
• Diversify energy sources: While the effects of renewable energy differ by region, sustained investment in renewable infrastructure, like electrified railroads and biofuel-compatible transportation systems, can facilitate sustainable mobility.
• Promote trade liberalization and industrial efficiency: The inverse correlation between trade openness and TCO₂ indicates that participation in global markets, coupled with effective logistics, might improve environmental outcomes. Advocating for sustainable industrial policies is equally essential.
• Augment regional collaboration: Landlocked nations ought to capitalize on regional trade accords and collaborative infrastructure initiatives to enhance connectivity, optimize border processes, and ease access to global markets.
• Advocate for technology and innovation: Digital technology for supply chain optimization, real-time tracking, and customs automation can markedly improve logistics efficiency while minimizing emissions.
• Capacity building and regulatory harmonization: Enhancing human capital and aligning customs and transportation rules with worldwide standards will facilitate effective and sustainable logistics systems.

This study possesses specific limitations that require attention in subsequent research. First, the study excludes landlocked nations in Africa and the Americas due to data availability, hence constraining the worldwide applicability of our conclusions. Subsequent studies should endeavor to include these regions to augment the thoroughness of the investigation. Secondly, essential governance-related issues, including political stability, institutional quality, and regulatory frameworks, are overlooked, despite their potential impact on policy execution and environmental results. Third, the study focuses on carbon dioxide emissions, neglecting other crucial environmental indicators such as nitrous oxide, sulfur dioxide, and ecological footprints, which are crucial for a comprehensive evaluation of environmental degradation. The analysis also focuses exclusively on road and rail freight transport, disregarding other pertinent modes such as air, inland waterways, and passenger transport, which could substantially influence emissions profiles. The absence of a comparison study with coastal states constrains the contextual comprehension of the distinct difficulties and opportunities encountered by LLCs. The dependence on annual data may obscure short-term variations and the direct effects of policy actions; hence, subsequent research should utilize higher-frequency or event-driven datasets. Minimizing these limitations will augment the profundity and pertinence of forthcoming studies and facilitate more efficacious, evidence-driven policymaking for sustainable transportation in LLCs.