The Impact of Port Total Factor Productivity on Carbon Dioxide Emissions in Port Cities: Evidence from the Yangtze River Ports

Abstract

This paper investigates the relationship between port productivity and carbon dioxide (CO₂) emissions in port cities. The study initially employs the global Malmquist productivity index (MPI) to measure productivity growth in 16 major inland ports along the Yangtze River, obtaining data on the ports’ total factor productivity (TFP). Through an analysis using the panel data model with two-way fixed effects, we find a positive correlation between the improvement of port TFP and the increase in CO₂ emissions in port cities. Further panel quantile regression analysis reveals the heterogeneity of this impact, especially in cities with medium and higher CO₂ emissions, where the positive effects of TFP on carbon emissions are particularly significant. The study also indicates a threshold effect of port size in the relationship between TFP and CO₂ emissions: in smaller ports, the impact of TFP improvement on CO₂ emissions is less significant; however, once the port size exceeds a certain threshold, the growth in TFP significantly promotes an increase in CO₂ emissions. These findings provide theoretical justification and decision-making references for policymakers to adopt effective measures to mitigate the growth of CO₂ emissions while promoting the efficiency of port production.

1. Introduction

The Yangtze River is China’s longest river and the third largest in the world, earning the reputation of the “Golden Waterway [1,2]”. In 2018, the annual cargo volume of the Yangtze River’s main course reached 2.69 billion tons, ranking it first in the world among inland rivers and making it the largest inland port system globally [3,4]. This achievement signifies that the Yangtze River’s water transport advantages have strengthened the economic and social connections in the Yangtze River Basin and have also become a focal area of attention for the Chinese government in promoting economic development [5]. Against this backdrop, enhancing the operational efficiency of inland ports along the Yangtze River has become crucial as ports are indispensable nodes in trade and supply chains [6]. In response, the Ministry of Transport of China released the “14th Five-Year” development plan for water transport in 2021, highlighting the current inadequacies in the specialization and intensive development levels of inland ports. The plan aspires to enhance the specialization and scaling of inland ports, focusing on major ports along the Yangtze River trunk line. These policies underscore the importance of improving the operational efficiency of inland ports, especially those along the Yangtze River.

However, the development of Yangtze River shipping is also accompanied by environmental pollution problems caused by CO₂ emissions and industrial activities along the river [4]. As a typical representative of basin economies, the Yangtze River Economic Belt concentrates on numerous key industries such as petrochemicals, steel, machinery, and automobiles. While these industries have propelled economic development, they have also become major sources of environmental pollution [7,8]. Particularly, the excessive consumption of fossil fuels has led to an overabundance of CO₂ emissions, placing immense pressure on the Yangtze River Economic Belt to achieve CO₂ reductions and affecting the realization of sustainable development [9,10]. In 2019, the CO₂ emissions from the Yangtze River Economic Belt accounted for more than 50% of the country’s total, posing a significant challenge to China’s “dual carbon” goals of reaching a carbon peak by 2030 and achieving carbon neutrality by 2060 [11,12].

Therefore, considering both the reduction in CO₂ emissions and the enhancement of port productivity has become a dual objective for policymakers. Understanding the interaction between the CO₂ emissions and port productivity of Yangtze River inland ports and their cities is of great importance for guiding policy formulation. The current research on carbon emissions in cities along the Yangtze River focuses on the extensive range of the Yangtze River Economic Belt, covering over 108 cities [13]. These studies explore the impact of factors such as construction land [12,14], industrialization [15], urbanization [16], foreign direct investment [16], economic growth [17], green innovation [13], industrial agglomeration [18,19], digital economy [20], and regional integration [21] on CO₂ emissions. However, due to the broad scope of the Yangtze River Economic Belt, most of the cities within it are not port cities, which leads to a general oversight in existing research on the potential impact of port activities on urban CO₂ emissions. In contrast, our study focuses on cities along the Yangtze River, thereby considering the ports’ role in filling the research gap in this field. Meanwhile, the existing literature on the impact of transportation infrastructure and its operational efficiency on urban CO₂ emissions, while considering the effects of roads, railways [22,23,24,25,26,27], and new logistics policies [28,29], often overlooks the crucial area of ports. Specifically, the total factor productivity (TFP) of ports, as a core indicator for measuring the production and service capacity of ports [3,30], is vital for ports and has become the focus of many studies [3]. In summary, we chose the TFP of ports along the Yangtze River to explore how it affects the CO₂ emissions of port cities for the following reasons: 1. Yangtze River ports, as economic and social linchpins, are crucial for the Yangtze River Basin, and their TFP is directly related to port operational efficiency, which could impact the CO₂ emissions levels of port cities. 2. The current literature on the impact of transportation infrastructure on urban CO₂ emissions lacks consideration of port impacts. 3. Reducing CO₂ emissions and enhancing port productivity are goals for policymakers, and our research aims to provide a basis for policymakers to formulate policies that can reduce CO₂ emissions and enhance port productivity simultaneously, supporting the coordinated development of these two goals.

The main contributions of this paper can be summarized in three aspects: 1. By supplementing research on the impact of transportation infrastructure and its operational efficiency on CO₂ emissions, this paper provides a new perspective for understanding the environmental impacts in the transportation sector. Unlike previous studies that focused on roads and railways, we have expanded our research subject to include ports for the first time and further analyzed the impact of port productivity on urban carbon emissions. 2. This study conducts an empirical analysis of the relationship between CO₂ emissions in cities along the Yangtze River and port activities, aiming to fill a significant gap in the existing literature: the impact of Yangtze River port activities on regional CO₂ emissions is often overlooked, and there is a lack of in-depth research on its effects. 3. By further revealing the heterogeneity in how port productivity affects CO₂ emissions in port cities, this paper provides an empirical foundation for governments to formulate comprehensive carbon reduction policy systems to balance improving port productivity with reducing CO₂ emissions in port cities.

The rest of this paper includes the following sections: Section 2 introduces our research framework. In Section 3, we review the literature related to the factors affecting CO₂ emissions in cities along the Yangtze River and the literature concerning the impact of transportation infrastructure on urban CO₂ emissions. Section 4 describes the research methods and data, including the calculation of port TFP and the panel data analysis process. Section 5 presents the results, including both the findings from the TFP calculations and the panel data analysis. Section 6 discusses these results. Section 7 concludes with conclusions and policy recommendations.

2. Research Framework

The STIRPAT model, proposed by Dietz and Rosa [31], is widely used in analyzing the driving factors behind environmental pressure. What sets this model apart is its ability not only to allow coefficients to be estimated in a parametric form but also to decompose and quantify the impacts of different factors, thus enabling it to adapt flexibly to various scenarios according to research needs [32]. Due to its broad applicability and flexibility, the STIRPAT model has become a popular method for studying the factors affecting carbon emissions [22].

The form of the STIRPAT model is as follows:

$$I=a{P}^b{A}^c{T}^{d}e$$

(1)

where $I$ represents the environmental impact; $P$, $A$, and $T$ respectively represent population size, affluence, and technological progress, which are considered key driving factors of CO₂ emissions; $b$, $c$, and $d$ are the exponents for the corresponding factors; $a$ is a constant term; and $e$ represents the error term. Xie, Fang, and Liu [22] improved the STIRPAT model by incorporating transportation infrastructure (such as roads) into the model to explore its impact on urban carbon emissions. The improved model adopts a logarithmic form:

$$lnI={\beta }_0+{\beta }_{1}lnROAD+{\beta }_{2}lnP+{\beta }_{3}lnA+{\beta }_{4}lnT+\epsilon$$

(2)

where $ROAD$ represents the scale of roads. Building on this, we further consider ports an essential component of transportation infrastructure and select the total factor productivity (TFP) of ports as a metric, since it can more comprehensively reflect the productivity and service capacity of ports, thereby more accurately assessing the impact of port activities on urban carbon emissions. Therefore, the model we construct is

$$lnI={\beta }_0+{\beta }_{1}TFP+{\beta }_{2}lnP+{\beta }_{3}lnA+{\beta }_{4}lnT+\epsilon$$

(3)

In this model, we did not take the logarithm of $TFP$ because it is already expressed as a percentage change, which inherently represents a relative change. This makes it suitable to be directly used as a variable in the model, allowing for a more direct reflection of the impact of port TFP on urban CO₂ emissions. We aim to determine the specific impact of port TFP (${\beta }_1$ coefficient) on urban CO₂ emissions. We constructed an econometric framework based on panel data from Yangtze River inland ports and their corresponding port cities to achieve this. Initially, we use the global Malmquist productivity index [33] to measure the TFP of major Yangtze River inland ports. Then, we integrate these TFP data into our constructed econometric panel model, aiming to delve into the relationship between port TFP and the CO₂ emissions of its city. To capture potential differences in this relationship across different levels of CO₂ emissions, we employed a two-way fixed effects and panel quantile model. Moreover, to consider the impact of varying port sizes, we also introduced a panel threshold model to analyze the impact of size heterogeneity on the relationship between port TFP and CO₂ emissions.

3. Literature Review

3.1. Factors Influencing CO₂ Emissions in Cities along the Yangtze River

Recent academic studies on CO₂ emissions in cities along the Yangtze River primarily focus on carbon emissions at the urban level within the Yangtze River Economic Belt (YREB), highlighting a range of determinants and impacts of CO₂ emissions.

Firstly, on the urban planning and economic dimensions, Wang, Wang, Wu, Yue, Wang, and Hu [12] and Liu, Xu, Wang, and Zhang [14] jointly examined the impact of urban construction land on CO₂ emissions in cities within the YREB. Wang, Wang, Wu, Yue, Wang, and Hu [12] concentrated on the impact of the scale of construction land, finding that an increase in the scale of construction land by 1% leads to a 0.3101% increase in CO₂ emissions. Liu, Xu, Wang, and Zhang [14] further discussed the effects of different categories of construction land, discovering that industrial and commercial lands are the main contributors to the increase in CO₂ emissions, while land associated with support facilities and infrastructure tends to mitigate CO₂ emissions. Wu and Zhang [16] initially calculated the carbon emission efficiency of cities in the YREB, then examined the influence of urbanization and foreign direct investment (FDI) on the efficiency of urban carbon emissions. They observed that urbanization leads to increased levels of carbon emissions, while FDI only causes increases in some areas. Li, Zhou, Xiao, Li, and Shan [17] analyzed the decoupling of economic growth from CO₂ emissions in cities along the Yangtze River Economic Belt, finding that some cities have achieved decoupling, meaning they have managed to reduce CO₂ emissions while maintaining economic growth.

Some studies have considered the factors affecting CO₂ emissions at the urban industrial level. For example, Zhang et al. [34] and Guo, Bai, Chen, and Luo [18] discussed the impact of the agglomeration of productive service industries and ICT industries on CO₂ emissions, respectively, confirming that both types of industry agglomeration significantly promote CO₂ emissions. Additionally, some research has focused on the impact of technological development levels. Fang, Gao, Wang, and Tian [13] and Ma, Hong, He, Luo, Liu, Zheng, Xia, and Xiao [20] investigated the effects of green innovation and the development of the digital economy on urban carbon emission efficiency, respectively. These studies suggest that green innovation and the development of the digital economy are key drivers in reducing carbon emissions. Policy interventions also play a crucial role, as demonstrated by the research of Fang et al. [35], Ai and Xu [21], and Li et al. [36], showing that green taxes, regional integration, and urban agglomeration policies have all contributed to the reduction in CO₂ emissions in cities.

Some studies have focused on the temporal and spatial variations in urban CO₂ emissions themselves, such as those by Lu et al. [37] and Chen et al. [38]. Lu, Lv, Wang, and Wei [37] found a declining trend in CO₂ emissions in the urban agglomerations of the Yangtze River Economic Belt. Chen, Tan, Mu, Wang, Chen, and He [38] focused on the synergistic relationship between pollution and carbon emissions, finding that the level of synergy between pollution control and carbon reduction in cities along the Yangtze River Economic Belt is on the rise. Lastly, some research has considered a comprehensive range of influencing factors, such as that conducted by Wang, Zhao, and Wang [2], who used the random forest machine learning algorithm to examine the relationship between ten influencing factors and carbon emissions.

Despite the comprehensive analysis provided by existing research, there is a noticeable lack of attention in the literature on the importance of Yangtze River port activities in the region’s CO₂ emissions.

3.2. The Impact of Transportation Infrastructure on Urban CO₂ Emissions

Current research on the relationship between transportation infrastructure and urban CO₂ emissions mainly focuses on roads and railways. The conclusions regarding whether the impact of transportation infrastructure on urban CO₂ emissions is positive or negative are inconsistent.

On one hand, some studies indicate that the development of transportation infrastructure leads to an increase in urban CO₂ emissions. For example, Xie, Fang, and Liu [22] used data from 283 Chinese cities between 2003 and 2013, with the length of highways as an indicator of transportation infrastructure, and found that the construction of transportation infrastructure increased carbon emissions, especially in larger cities. This phenomenon may stem from the urban economic growth and technological innovation promoted by transportation infrastructure. Similarly, Xie, Fang, and Liu [25], by analyzing the per capita road area in cities, also found that transportation infrastructure has a negative impact on the urban environment. Zeng, Chai, Stringer, Li, Wang, Deng, Ma, and Ren [26] studied the passenger and freight flow on railways and highways, as well as railway length, from the perspective of regional spatial spillover, considering these as significant contributors to urban carbon emissions. Meanwhile, Xiao, Pang, Yan, Kong, and Yang [24] pointed out that road infrastructure, through the effect of industrial agglomeration, has exacerbated carbon emissions in Chinese prefecture-level cities.

On the other hand, there are studies that believe the development of transportation infrastructure can help reduce urban CO₂ emission levels. Li and Luo [27], through the study of urban road areas, found that improvements in transportation infrastructure significantly reduced the level of urban carbon emissions. Wang et al. [39], when considering the impact of transportation factors in China’s megacities on CO₂ emissions, also found that an increase in urban road density helps to reduce carbon emissions. In terms of railways, Sun and Li [23] regarded the operation of urban high-speed rail as a “quasi-natural experiment” and found that the operation of high-speed rail significantly reduced CO₂ emissions in cities along the line.

In addition to studies directly focused on transportation infrastructure itself, there has been research considering policies to improve the operational efficiency of transportation infrastructure, assessing the impact of these policies on urban CO₂ emissions. For example, Lu, Xiao, Jiao, Du, and Huang [29] and Pan, Li, Wang, Zong, and Song [28] studied the implementation of smart transportation policies and smart logistics policies, respectively, finding that both policies effectively curbed urban CO₂ emissions.

Although existing research has extensively discussed the impact of roads and railways as transportation infrastructures on urban CO₂ emissions, these studies often overlook the critical role of ports in this process. Our research aims to fill this gap, exploring the relationship between port productivity and CO₂ emissions in port cities.

4. Methodology and Data

4.1. Calculation of Total Factor Productivity for Yangtze River Inland Ports

Total factor productivity (TFP) is a crucial indicator for measuring port productivity, reflecting the rate of change in total output relative to total inputs [3,40,41]. TFP aggregates multiple input (x) and output (y) indicators to measure or decompose changes in port productivity over time or among different ports [42]. In this regard, the Malmquist productivity index (MPI) is an effective tool for measuring changes in port TFP between two time periods and decomposing the sources of efficiency change [43]. The Malmquist TFP index measures TFP changes by comparing the distance of each data point relative to a common technology, with its main advantage being that it does not rely on input and output prices or related market clearing assumptions [40,42].

While retaining the advantages of the traditional Malmquist productivity index, we have adopted the more advanced global Malmquist productivity index as a tool for measuring changes in the productivity of Yangtze River inland ports. This index, developed by Pastor and Lovell [33], not only meets the circularity assumption but also avoids the infeasibility of mixed-period models that may arise in the traditional Malmquist productivity index [44,45].

Specifically, the formula for measuring TFP changes using the global Malmquist productivity index is as follows:

$$TFPCH=\frac{{TFP}_i^{t+1}}{{TFP}_i^t}=M_i^G\left(y_{i,t+1},x_{i,t+1},y_{i,t},x_{i,t}\right)=\frac{D_i^G\left(y_{i,t+1},x_{i,t+1}\right)}{D_i^G\left(y_{i,t},x_{i,t}\right)}$$

(4)

where variables x and y respectively represent the input vector and the output vector of the port. $D_i^G$ is the output-oriented distance function based on global reference technology. t and t + 1 represent two time periods. $M_i^G$ can be used to measure the TFP change in port $i$ from period t to t + 1. $M_i^G>1$ indicates an increase in TFP for Yangtze River inland port $i$, while $M_i^G=1$ and $M_i^G<1$ respectively denote no change and a decrease in TFP for Yangtze River inland port $i$. Under the VRS (variable returns to scale) condition, the global MPI can be decomposed in three directions using the method of Ray and Desli [46], with the decomposition process as follows:

$$\begin{array}{c}{M}_i^G\left(y_{i,t+1},x_{i,t+1},y_{i,t},x_{i,t}\right)=\frac{D_i^G\left(y_{i,t+1},x_{i,t+1}\right) \times SE\left(y_{i,t+1},x_{i,t+1}\right)}{D_i^G\left(y_{i,t},x_{i,t}\right) \times SE\left(y_{i,t},x_{i,t}\right)}\hfill \\ =\frac{D_i^{t+1}\left(y_{i,t+1}, x_{i,t+1}\right)}{D_i^t\left(y_{i,t}, x_{i,t}\right)}\times \left[\frac{\frac{D_i^G\left(y_{i,t+1}, x_{i,t+1}\right)}{D_i^{t+1}\left(y_{i,t+1}, x_{i,t+1}\right)}}{\frac{D_i^G\left(y_{i,t}, x_{i,t}\right)}{D_i^t\left(y_{i,t}, x_{i,t}\right)}}\right]\times \frac{SE\left(y_{i,t+1}, x_{i,t+1}\right)}{SE\left(y_{i,t}, x_{i,t}\right)}\hfill \\ ={TEC}_{i,(t,t+1)}\times \left[\frac{{BPG}_{i,t+1}^G(y_{i,t+1},x_{i,t+1})}{{BPG}_{i,t}^G(y_{i,t},x_{i,t})}\right]\times {SEC}_{i,\left(t,t+1\right)}={TEC}_{i,(t,t+1)}\times {BPC}_{i,\left(t,t+1\right)}\times {SEC}_{i,\left(t,t+1\right)}\hfill \end{array}$$

(5)

To explore the sources of productivity changes in Yangtze River ports, as demonstrated in Equation (5), we decomposed port productivity changes under the assumption of variable returns to scale (VRS) into three parts: pure technical efficiency change (TEC), best practice change (BPC), and scale efficiency changes (SECs) [45]. Specifically, a TEC value greater than 1 indicates an improvement in the port’s technical efficiency, whereas a value less than 1 suggests a decline in technical efficiency. The BPC value reflects how closely port technology aligns with global technology development, where a BPC value greater than 1 indicates that the port’s technology has become closer to the global technology level during the comparison period, and a value less than 1 indicates that the port’s technology has moved away from the global technology level. The SEC value indicates whether changes in the port’s scale have contributed to productivity growth, with a value greater than 1 signifying an improvement in scale efficiency and a value less than 1 indicating a reduction in scale efficiency.

In conducting this analysis, we utilized the data envelopment analysis (DEA) method to calculate output distance functions. DEA is a non-parametric technique that evaluates the performance of a set of homogeneous decision-making units (DMUs) with multiple inputs and outputs by creating efficiency frontiers based on a piecewise linear surface [43,47]. In studies related to port TFP, the use of DEA to calculate distance functions and thereby construct the Malmquist TFP index has been widely applied [6,30,40,42,48,49,50]. For more technical details on the global Malmquist productivity index based on DEA, please refer to the relevant literature by Pastor and Lovell [33].

In our study of the TFP of Yangtze River inland ports, we selected specific input and output variables based on the analysis of the early literature and the availability of data. Song and Liu [3] argued that using cargo throughput and container throughput as output indicators can accurately reflect the production objectives of Yangtze River inland ports. Following this logic, we also adopted these output indicators. At the same time, considering that the number and length of berths are direct determinants of quay transfer operation efficiency and are also key indicators of port production and service capacity [51], we decided to use the number and length of productive berths at Yangtze River inland ports as the main input indicators. The descriptive statistics of these indicators are presented in

Table 1. Descriptive statistics of input and output variables.

$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{i}\mathbf{a}\mathbf{b}\mathbf{l}\mathbf{e}$	$\mathbf{O}\mathbf{b}\mathbf{s}$	$\mathbf{M}\mathbf{e}\mathbf{a}\mathbf{n}$	$\mathbf{M}\mathbf{a}\mathbf{x}$	$\mathbf{M}\mathbf{i}\mathbf{n}$	$\mathbf{S}\mathbf{t}\mathbf{d}.\mathbf{D}\mathbf{e}\mathbf{v}.$
$\mathrm{O}\mathrm{u}\mathrm{t}\mathrm{p}\mathrm{u}\mathrm{t}\mathrm{V}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{b}\mathrm{l}\mathrm{e}$
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{a}\mathrm{i}\mathrm{n}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{g}\mathrm{h}\mathrm{p}\mathrm{u}\mathrm{t}$ $\left(\mathrm{10,000}\mathrm{T}\mathrm{E}\mathrm{U}\right)$	$176$	$47.26675$	$330.54$	$0.4$	$66.4993$
$\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{g}\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{o}\mathrm{u}\mathrm{g}\mathrm{h}\mathrm{p}\mathrm{u}\mathrm{t}$ $\left(\mathrm{10,000}\mathrm{t}\right)$	$176$	$9517.479$	$\mathrm{33,620.35}$	$554$	$6897.967$
$\mathrm{I}\mathrm{n}\mathrm{p}\mathrm{u}\mathrm{t}\mathrm{V}\mathrm{a}\mathrm{r}\mathrm{i}\mathrm{a}\mathrm{b}\mathrm{l}\mathrm{e}$
$\mathrm{N}\mathrm{u}\mathrm{m}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{o}\mathrm{f}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{s}\mathrm{u}\mathrm{s}\mathrm{e}\mathrm{d}$ $\mathrm{i}\mathrm{n}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}$	$176$	$166.8523$	$15$	$880$	$177.8984$
$\mathrm{L}\mathrm{e}\mathrm{n}\mathrm{g}\mathrm{t}\mathrm{h}\mathrm{o}\mathrm{f}\mathrm{b}\mathrm{e}\mathrm{r}\mathrm{t}\mathrm{h}\mathrm{s}\mathrm{u}\mathrm{s}\mathrm{e}\mathrm{d}$ $\mathrm{i}\mathrm{n}\mathrm{p}\mathrm{o}\mathrm{r}\mathrm{t}\mathrm{p}\mathrm{r}\mathrm{o}\mathrm{d}\mathrm{u}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{n}\left(\mathrm{m}\right)$	$176$	$\mathrm{16,366.36}$	$76017$	$2631$	$\mathrm{15,666.94}$

.

The port sample includes 16 major inland ports in the Yangtze River Basin from 2009 to 2019. Introduced in order from upstream to downstream, the specific list of these 16 ports is as follows: Chongqing, Yichang, Wuhan, Huangshi, Jiujiang, Anqing, Chizhou, Tongling, Wuhu, Ma’anshan, Nanjing, Zhenjiang, Yangzhou, Taizhou, Changzhou, and Nantong. The primary considerations for selecting these specific ports were the availability of data and the representativeness of these ports. This sample scope is similar to the research by Song and Liu [3], which explored the impact of internet development levels on the TFP of Yangtze River ports. This further confirms the representativeness and applicability of our chosen port sample. The port data used in this paper are all sourced from the China Statistical Yearbook, China Port Statistics Yearbook, and the Shanghai International Shipping Institute.

4.2. Econometric Models

4.2.1. Two-Way Fixed Effects Model and Panel Quantile Model

We employed panel data models to study the relationship between the TFP of ports along the Yangtze River and the CO₂ emissions of port cities, with its simplified form as follows:

$$Y_{i,t}={\beta }_0+{\beta }_1{TFP}_{i,t}+{\beta }_2{X}_{i,t}+YearFE+CityFE+{\epsilon }_{i,t}$$

(6)

where $i$ and $t$ respectively represent the port city and the year, $Y_{i,t}$ is the dependent variable (city CO₂ emissions), $X_{i,t}$ is a vector of control variables, ${\epsilon }_{i,t}$ represents the random error, $CityFE$ denotes city fixed effects, which control for the impact of static city characteristics (such as geographical location, etc.) on CO₂ emissions in port cities, and $YearFE$ represents time fixed effects, which can control for national or global trends and events affecting all port cities, such as changes in national environmental policies and fluctuations in international oil prices.

When dealing with environmental data, we often encounter challenges, including distinct peaks or fat tails in the data, which implies unequal variation within the data distribution [52]. Addressing this challenge, the traditional two-way fixed effects (TWFE) ordinary least squares (OLS) estimation method may not provide a sufficient solution, as it primarily focuses on the conditional expectations of the dependent variable (i.e., the mean) [53,54]. This is particularly relevant when analyzing the relationship between port TFP and CO₂ emissions from port cities, where a method capable of capturing the differences in relationships at various emission levels is deemed necessary. For this reason, we opt for panel quantile regression analysis due to the following considerations: 1. quantile regression techniques offer an effective solution by allowing coefficients to vary with different quantiles, enabling a detailed understanding of how the relationship between port TFP and CO₂ emissions changes with different emission levels [55]. 2. Compared to traditional OLS estimation methods, quantile regression provides a more robust estimation technique. It does not impose strict assumptions on the error term, allowing the method to handle potential issues of heteroscedasticity, outliers, and unobserved heterogeneity more effectively, which are factors that could significantly impact the accuracy of estimations [52,56].

The form of the panel quantile model is as follows:

$$Q_{Y_{i,t}}^{\left(\tau \right)}\left(\tau |X_{i,t}\right)={\alpha }_i+{\beta }_{\tau }{X}_{i,t}+u_{i,t}i=1,2,\dots ,N,t=\mathrm{1,2},\dots ,T$$

(7)

where $Y_{i,t}$ and $X_{i,t}$ are the dependent variable (city CO₂ emissions) and the vector of explanatory variables (port TFP and other control variables), respectively. $Q_{Y_{i,t}}^{\left(\tau \right)}\left(\tau |X_{i,t}\right)$ represents the ${\tau }^{th}$ conditional quantile of the dependent variable, with $\tau \in \left(\mathrm{0,1}\right)$. ${\beta }_{\tau }$ represents the regression coefficient estimate of the explanatory variables at the ${\tau }^{th}$ quantile, which is our primary focus. We assign the quartiles of $\tau$ as 0.25, 0.5, and 0.75. The 25th percentile (Q25) is used to illustrate the impact of port TFP on cities with lower CO₂ emissions, while the 50th percentile (Q50) and the 75th percentile (Q75) represent cities with medium and higher CO₂ emissions, respectively. To obtain the parameter estimates for ${\mathsf{\beta }}_{\mathsf{\tau }}$, the following problem needs to be solved:

$$\underset{\left(\alpha ,\beta \right)}{\min }\sum _{k=1}^K\sum _{t=1}^T\sum _{i=1}^N{\omega }_k{\rho }_{{\tau }_k}\left(Y_{i,t}-{\alpha }_i-{\beta }_{\tau }{X}_{i,t}\right)+\lambda \sum _{i=1}^N\left|{\alpha }_i\right|$$

(8)

where $k$ is the index of quantiles, ${\rho }_{{\tau }_k}$ is the loss function, $\lambda$ is the penalty factor, and ${\omega }_k$ represents the contribution of the $k^{th}$ quantile to the fixed effects.

4.2.2. Panel Threshold Model

This paper posits that the impact of port productivity on CO₂ emissions in port cities varies with the size of the port, necessitating a consideration of the presence of threshold effects. To delve deeper into this heterogeneity, we employed the threshold panel model developed by Hansen, which can capture jumps or structural changes in the relationships between different variables [57]. The so-called threshold effect refers to a change in the direction or magnitude of another economic variable when a certain economic indicator exceeds a specific value, which is termed the threshold [57]. By endogenously dividing different port size intervals based on the characteristics of port size data, we further explored how port TFP affects CO₂ emissions in port cities across different port size intervals.

Here, we primarily introduce the setup of the single-threshold model used, detailed as follows:

$$Y_{i,t}={\delta }_i+{\beta }^{\prime }{X}_{i,t}+{\theta }_1{g}_{i,t}I\left(d_{i,t}\le r\right)+{\theta }_2{g}_{i,t}I\left(d_{i,t}>r\right)+{\omega }_{i,t}$$

(9)

where $Y_{i,t}$ still represents the dependent variable (CO₂ emissions of port cities). $X_{i,t}$ is the set of control variables, with ${\beta }^{\prime }$ as their corresponding coefficients. $r$ is a specific threshold (a specific port size). $d_{i,t}$ is the threshold variable (port size), for which we use the length of the port’s production berths (in meters) as the representative variable for port size, and $I$ is an indicator function. ${\delta }_i$ reflects the individual impact of each port city. The random error term ${\omega }_{i,t}$ is assumed to be independently and identically distributed. The estimation is based on the principle of minimizing the sum of squared residuals (SSR). Furthermore, it is necessary to verify the existence of a threshold effect, i.e., to test whether the model estimates of ${\theta }_1$ and ${\theta }_2$ are significantly different for the two groups of samples divided by the threshold value. We use the bootstrap method to obtain the asymptotic distribution and thereby derive the corresponding probability p-values and confidence intervals [58].

4.2.3. Variables and Data

The dependent variable of this study is the CO₂ emissions (CO₂) of port cities along the Yangtze River. The data on CO₂ emissions for port cities along the Yangtze River are derived from the CEADs database by Shan et al. [59], considered the most comprehensive dataset for estimating carbon emissions at the urban level in China [60]. Following the approach of Gao et al. [61], we extract the CO₂ emissions data for port cities along the Yangtze River from the CEADs database and take the logarithm to represent the level of CO₂ emissions of port cities along the Yangtze River.

The core explanatory variable is port TFP ($TFP$). To calculate TFP, we follow the method proposed by Song and Liu [3], initially using the global Malmquist index to estimate the TFP growth rate of each port from 2009 to 2019. Subsequently, taking the total factor productivity of 2009 as the base value (set to 1), we accumulate the TFP growth rates over this period to obtain the TFP index for each port.

Via the STIRPAT model, we have selected the following variables as control variables: (1) technological progress ($T$): the natural logarithm of the number of patent inventions, which reflects the impact of technological advancement [62]. (2) Population ($P$): The natural logarithm of the population size. Population is considered a key factor influencing urban CO₂ emissions. (3) Affluence ($A$): we measure affluence using the natural logarithm of per capita GDP, which is another key factor affecting urban CO₂ emissions [63].

Table 2. Descriptive statistics of variables for the panel data model.

$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{i}\mathbf{a}\mathbf{b}\mathbf{l}\mathbf{e}$	$\mathbf{O}\mathbf{b}\mathbf{s}.$	$\mathbf{M}\mathbf{e}\mathbf{a}\mathbf{n}$	$\mathbf{S}\mathbf{t}\mathbf{d}.\mathbf{D}\mathbf{e}\mathbf{v}.$	$\mathbf{M}\mathbf{i}\mathbf{n}$	$\mathbf{M}\mathbf{a}\mathbf{x}$
CO2	$176$	$3.759$	$0.715$	$1.927$	$5.184$
$TFP$	$176$	$1.579$	$0.744$	$0.541$	$4.909$
$P$	$176$	$5.091$	$0.979$	$3.802$	$7.816$
$A$	$176$	$11.434$	$0.464$	$9.865$	$12.249$
$T$	$176$	$6.110$	$1.801$	$0.693$	$9.422$

presents the descriptive statistics of the variables. Apart from the port TFP data, which the authors calculated, all other city-level data were obtained from the China City Statistical Yearbook, the China Statistical Yearbook for Regional Economy, and the statistical yearbooks of various port cities.

5. Results

5.1. Total Factor Productivity of Yangtze River Inland Ports

Table 3. MPI summary of port means (2009–2019).

$\mathbf{P}\mathbf{O}\mathbf{R}\mathbf{T}$	$\mathbf{T}\mathbf{F}\mathbf{P}\mathbf{C}\mathbf{H}$	$\mathbf{T}\mathbf{E}\mathbf{C}$	$\mathbf{B}\mathbf{P}\mathbf{C}$	$\mathbf{S}\mathbf{E}\mathbf{C}$
$\mathrm{C}\mathrm{h}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{q}\mathrm{i}\mathrm{n}\mathrm{g}$	$1.107$	$0.989$	$1.097$	$1.027$
$\mathrm{Y}\mathrm{i}\mathrm{c}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{g}$	$1.054$	$1.011$	$1.033$	$1.172$
$\mathrm{W}\mathrm{u}\mathrm{h}\mathrm{a}\mathrm{n}$	$1.155$	$1.057$	$1.098$	$1.000$
$\mathrm{H}\mathrm{u}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{s}\mathrm{h}\mathrm{i}$	$1.181$	$1.144$	$1.049$	$0.990$
$\mathrm{J}\mathrm{i}\mathrm{u}\mathrm{j}\mathrm{i}\mathrm{a}\mathrm{n}\mathrm{g}$	$1.182$	$1.101$	$1.075$	$1.001$
$\mathrm{A}\mathrm{n}\mathrm{q}\mathrm{i}\mathrm{n}\mathrm{g}$	$1.086$	$1.030$	$1.055$	$0.998$
$\mathrm{C}\mathrm{h}\mathrm{i}\mathrm{z}\mathrm{h}\mathrm{o}\mathrm{u}$	$1.169$	$1.115$	$1.053$	$1.000$
$\mathrm{T}\mathrm{o}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{i}\mathrm{n}\mathrm{g}$	$1.076$	$1.026$	$1.043$	$1.010$
$\mathrm{W}\mathrm{u}\mathrm{h}\mathrm{u}$	$1.098$	$1.035$	$1.059$	$1.003$
$\mathrm{M}\mathrm{a}\mathrm{’}\mathrm{a}\mathrm{n}\mathrm{s}\mathrm{h}\mathrm{a}\mathrm{n}$	$1.061$	$1.011$	$1.048$	$1.006$
$\mathrm{N}\mathrm{a}\mathrm{n}\mathrm{j}\mathrm{i}\mathrm{n}\mathrm{g}$	$1.098$	$1.000$	$1.090$	$1.010$
$\mathrm{Z}\mathrm{h}\mathrm{e}\mathrm{n}\mathrm{j}\mathrm{i}\mathrm{a}\mathrm{n}\mathrm{g}$	$1.122$	$1.050$	$1.072$	$0.987$
$\mathrm{Y}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{z}\mathrm{h}\mathrm{o}\mathrm{u}$	$1.052$	$1.000$	$1.044$	$1.011$
$\mathrm{T}\mathrm{a}\mathrm{i}\mathrm{z}\mathrm{h}\mathrm{o}\mathrm{u}$	$1.061$	$1.000$	$1.059$	$1.003$
$\mathrm{C}\mathrm{h}\mathrm{a}\mathrm{n}\mathrm{g}\mathrm{z}\mathrm{h}\mathrm{o}\mathrm{u}$	$1.048$	$1.000$	$1.032$	$1.040$
$\mathrm{N}\mathrm{a}\mathrm{n}\mathrm{t}\mathrm{o}\mathrm{n}\mathrm{g}$	$1.066$	$1.000$	$1.064$	$1.002$
$\mathrm{M}\mathrm{e}\mathrm{a}\mathrm{n}$	$1.101$	$1.034$	$1.061$	$1.016$

shows the average total factor productivity (TFP) growth rates for 16 Yangtze River inland ports from 2009 to 2019. Overall, we observe an improvement in the TFP of the 16 Yangtze River inland ports from 2009 to 2019, with an average annual TFP growth rate of 10.1% (mean TFPCH = 1.101). The average annual growth rates for technological efficiency change (TEC), best practice change (BPC), and scale efficiency change (SEC) were 3.4%, 6.1%, and 1.6%, respectively, all contributing positively to TFP growth, with BPC having the most significant impact. This finding indicates that the growth in port TFP during the observation period can primarily be attributed to improvements in technological levels, i.e., the gradual convergence of port technology levels toward the global standard, which has driven enhancements in port operational efficiency.

5.2. Baseline Regression Results

The results obtained from the two-way fixed effects panel regression are shown in

Table 4. Baseline regression results.

$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{i}\mathbf{a}\mathbf{b}\mathbf{l}\mathbf{e}\mathbf{s}$	$\mathbf{T}\mathbf{W}\mathbf{F}\mathbf{E}$	$\mathbf{Q}25$	$\mathbf{Q}50$	$\mathbf{Q}75$
$TFP$	0.078 *** (0.022)	$0.020$ $\left(0.031\right)$	0.039 * (0.022)	0.055 * (0.030)
$P$	$0.074$ $\left(0.097\right)$	$-0.116$ $\left(0.101\right)$	$-0.091$ $\left(0.073\right)$	$-0.071$ $\left(0.098\right)$
$A$	0.351 *** (0.108)	0.217 ** (0.101)	0.155 ** (0.073)	$0.104$ $\left(0.097\right)$
$T$	0.141 *** (0.027)	0.070 ** (0.031)	0.090 *** (0.022)	0.106 *** (0.030)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{n}\mathrm{t}$	$-1.414$ $\left(1.606\right)$	-	-	-
$F$	$19.87$	-	-	-
$N$	$176$	$176$	$176$	$176$
$R^2$	$0.688$	-	-	-

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

(TWFE). Specifically, a 1% increase in port TFP leads to a 0.054% increase in CO₂ emissions from port cities. This means that the productivity improvements in the Yangtze River ports from 2009 to 2019 have increased CO₂ emissions from the corresponding port cities.

We further employed the panel quantile model to ascertain the magnitude and statistical significance of the TFP coefficient at the Q25, Q50, and Q75 quantile levels. The impact coefficient of TFP on CO₂ emissions passed the significance test at the Q50 and Q75 quantile levels, with coefficients of 0.048 and 0.058, respectively. This means that a 1% increase in port TFP will lead to a 0.048% increase in CO₂ emissions for port cities with medium levels of CO₂ emissions, while the effect is more pronounced for port cities with higher levels of CO₂ emissions, resulting in a 0.058% increase. However, the impact coefficient of TFP on CO₂ emissions did not pass the significance test at the Q25 quantile level, indicating that changes in port TFP do not affect port cities with lower levels of CO₂ emissions.

5.3. Threshold Effect Analysis

We first constructed the LR statistic and utilized the bootstrap method with 300 repetitions to determine the number of thresholds. As shown in

Table 5. Threshold existence test.

$\mathbf{T}\mathbf{h}\mathbf{r}\mathbf{e}\mathbf{s}\mathbf{h}\mathbf{o}\mathbf{l}\mathbf{d}$	$\mathbf{H}\mathbf{y}\mathbf{p}\mathbf{o}\mathbf{t}\mathbf{h}\mathbf{e}\mathbf{s}\mathbf{i}\mathbf{s}$	$\mathbf{R}\mathbf{S}\mathbf{S}$	$\mathbf{M}\mathbf{S}\mathbf{E}$	$\mathbf{F}\mathbf{s}\mathbf{t}\mathbf{a}\mathbf{t}$	$\mathbf{P}\mathbf{r}\mathbf{o}\mathbf{b}$
$\mathrm{S}\mathrm{i}\mathrm{n}\mathrm{g}\mathrm{l}\mathrm{e}$	$\mathrm{H}0:\mathrm{n}\mathrm{o}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}$ $\mathrm{H}1:\mathrm{o}\mathrm{n}\mathrm{e}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}$	$2.019$	$0.012$	12.88 *	$0.08$
$\mathrm{D}\mathrm{o}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}$	$\mathrm{H}0:\mathrm{o}\mathrm{n}\mathrm{e}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}$ $\mathrm{H}1:\mathrm{d}\mathrm{o}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}$	$2.015$	$0.012$	$0.31$	$0.98$
$\mathrm{T}\mathrm{r}\mathrm{i}\mathrm{p}\mathrm{l}\mathrm{e}$	$\mathrm{H}0:\mathrm{d}\mathrm{o}\mathrm{u}\mathrm{b}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}$ $\mathrm{H}1:\mathrm{t}\mathrm{r}\mathrm{i}\mathrm{p}\mathrm{l}\mathrm{e}\mathrm{t}\mathrm{h}\mathrm{r}\mathrm{e}\mathrm{s}\mathrm{h}\mathrm{o}\mathrm{l}\mathrm{d}$	$1.999$	$0.012$	$1.34$	$0.69$

Note: * p < 0.1.

, the statistical significance of the first threshold is below 0.1, but the second and third thresholds are not statistically significant. This indicates that when port size is considered the threshold variable, the impact of port TFP on CO₂ emissions from port cities exhibits a significant single-threshold effect. Estimating the single-threshold panel model, the obtained single threshold is 7867, with a corresponding 95% confidence interval of [7854, 7889].

According to the estimation results in

Table 6. Threshold regression results.

$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{i}\mathbf{a}\mathbf{b}\mathbf{l}\mathbf{e}\mathbf{s}$	$\mathbf{T}\mathbf{h}\mathbf{r}\mathbf{e}\mathbf{s}\mathbf{h}\mathbf{o}\mathbf{l}\mathbf{d}\mathbf{M}\mathbf{o}\mathbf{d}\mathbf{e}\mathbf{l}$
$TFP{(d}_{i,t}\le 7867)$	$-0.008$ $\left(0.024\right)$
$TFP(d_{i,t}>7867)$	0.062 *** (0.021)
$P$	−0.121 * (0.064)
$A$	0.174 *** (0.059)
$T$	0.083 *** (0.018)
$Constant$	$1.815$ $\left(0.736\right)$
$F$	$49.78$
$N$	$176$
$R^2$	$0.164$

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

, we observe a significant threshold effect of port size on the relationship between the TFP and CO₂ emissions in port cities. Specifically, when the port size is below the threshold, the impact of port TFP on CO₂ emissions is negligible, with a coefficient of only −0.008, which lacks statistical significance. Conversely, when the port size exceeds this threshold, the positive impact of port TFP on CO₂ emissions significantly strengthens, with the coefficient reaching 0.062, and it is statistically significant at the 1% level. This finding reveals an important phenomenon: in smaller ports, improvements in TFP hardly affect CO₂ emissions, whereas for larger ports, an increase in TFP significantly promotes CO₂ emissions in port cities.

6. Discussion

The results reveal several intriguing findings. Firstly, the TFP of ports has a positive and significant impact on the CO₂ emissions of port cities, consistent with the research outcomes of Xie, Fang, and Liu [22], Xie, Fang, and Liu [25], and Zeng, Chai, Stringer, Li, Wang, Deng, Ma, and Ren [26]. Their studies also confirmed that improving transportation infrastructure contributes to the growth of urban CO₂ emissions. However, this study is the first to propose this perspective from the standpoint of port infrastructure. As critical nodes in the global supply chain, ports are not only logistics hubs but also significant sources of energy consumption, environmental pollution, and CO₂ emissions that contribute to climate change [64]. Ports themselves generate pollution, primarily from dust and exhaust emissions during loading and unloading operations and from fuel combustion by ships. As the core of maritime shipping, the high fuel consumption and harmful emissions of ships pose a significant challenge to the environment, affecting air quality and public health in marine, coastal, and even inland regions [65,66]. Furthermore, we argue that the enhancement of port production efficiency represents a higher volume of goods handled and promotes the expansion of logistics and related industrial chain activities, increasing the demand for energy sources such as oil and coal. This, in turn, significantly raises the CO₂ emissions of port cities.

Secondly, the impact of port TFP on CO₂ emissions intensifies with the increase in CO₂ emission levels, indicating that the effect of port TFP on CO₂ emissions is particularly significant in high-pollution areas. Some studies have shown that improvements in transportation infrastructure can significantly enhance the production efficiency of enterprises. In this regard, enterprises in high-pollution regions tend to have more robust CO₂ emission capabilities, meaning that while improving port efficiency increases these enterprises’ production efficiency, it does not effectively reduce their CO₂ emissions. On the contrary, this enhancement gives enterprises a greater production capacity, which may lead to more pollution. Therefore, this finding provides essential implications for policymaking. If policies overly focus on enhancing port efficiency while neglecting the importance of reducing CO₂ emissions, they may exacerbate environmental pressures. This could have a more severe environmental impact than if port efficiency had not been improved. Hence, when formulating policies, there needs to be a balanced consideration of the relationship between improving port efficiency and reducing CO₂ emissions to ensure a better balance between economic development and environmental protection.

Thirdly, the study indicates that the port’s size significantly influences port TFP’s impact on CO₂ emissions. Specifically, only when the size of a port—measured by the length of its production berths—exceeds 7867 m does the enhancement of port TFP lead to an increase in the CO₂ emissions of the city in which it is located. This phenomenon reveals an important point: as the size of a port increases, the improvement in port productivity not only facilitates the transportation of more goods but also amplifies the impact of port productivity enhancement on CO₂ emissions. Therefore, for smaller ports, the impact of productivity enhancement on urban CO₂ emissions is relatively minor and should not be a primary consideration. However, for larger ports, such as those in Chizhou, Jiujiang, Ma’anshan, Nanjing, Nantong, Taizhou, Wuhu, Wuhan, Yichang, Zhenjiang, and Chongqing, as of 2019, reducing CO₂ emissions should become a focal issue. This means policymaking should encourage small-scale ports to develop through expansion and efficiency improvement. At the same time, more stringent regulatory measures should be applied to large-scale ports to promote the reduction in CO₂ emissions.

7. Conclusions

This study analyzed the total factor productivity (TFP) of the Yangtze River inland ports from 2009 to 2019 and its impact on port cities’ carbon emissions to reveal how improvements in port production efficiency affect CO₂ emissions in port cities. The results show that the overall improvement in the TFP of the Yangtze River inland ports reflects a significant increase in operational efficiency, primarily due to technological progress. However, analyses using the panel data model with two-way fixed effects indicate that improved port production efficiency has concurrently led to increased CO₂ emissions in port cities. Further investigation through panel quantile regression analysis reveals that the impact of TFP on CO₂ emissions varies among port cities with different emission levels and is significantly more pronounced in cities with medium and higher CO₂ emission levels. Moreover, this study also identifies a significant threshold effect of port size on the relationship between TFP and CO₂ emissions: in smaller ports, improvements in TFP have a negligible impact on CO₂ emissions; however, once the scale exceeds a certain threshold, the enhancement in TFP significantly promotes an increase in CO₂ emissions.

In light of these findings, it is recommended that policymakers promote the improvement of port production efficiency while simultaneously adopting measures to mitigate the growth of carbon emissions. Specific measures include promoting the application of green port technologies, enhancing energy efficiency management, encouraging the use of clean energy, and optimizing the logistics chain between ports and cities to reduce carbon emissions. For larger ports, given their greater potential environmental impact, more stringent environmental policies and standards should be developed and enforced to ensure that port development does not come at the expense of environmental quality.

The contribution of this paper is that it offers a new perspective on understanding the environmental impacts of transportation infrastructure, particularly port activities. It fills a research void concerning the relationship between carbon emissions and port activities in cities along the Yangtze River. It provides an empirical foundation for developing policies to balance improving port productivity with reducing carbon emissions.

This study is somewhat limited, mainly due to the lack of detailed data on ports, such as the specific number of pieces of equipment, yard areas, and the number of employees. The absence of these detailed data is due to difficulties in data collection, which has led us to rely on traditional input variables to calculate port TFP, thus limiting our research to some extent.

Future research should consider a more comprehensive analysis of the impact of the internal interactions within transportation infrastructure on urban carbon emissions. This includes the role of individual transportation infrastructures like roads, railways, and ports and the exploration of their synergistic effects. Moreover, future studies could extend to coastal ports in China, further exploring the mechanisms and strategies affecting them to balance the enhancement of port production efficiency with reducing carbon emissions.