The dynamic relationships among growth in the logistics industry, energy consumption, and carbon emission: recent evidence from China

Abstract

In the post-pandemic era, there is a need to achieve the dynamic and coordinated development of growth in the logistics industry, energy consumption, and carbon dioxide (CO₂) emissions in China’s four major economic regions to ensure the high-quality development of green logistics in China. Using the data indicators of growth in the logistics industry, energy consumption, CO₂ emissions, and fixed asset investment in 30 Chinese provinces from 2004 to 2018, a panel vector autoregressive model was established for the four major economic development regions of central, east, west, and northeast China. The model coefficients were estimated using the systematic generalized matrix estimation method (System-GMM), which was evaluated by a Granger causality test. The model coefficients were estimated using the System-GMM method, and the dynamic relationships between growth in the logistics industry, energy consumption, and CO₂ emissions was obtained through a Granger causality test, impulse response analysis, and variance decomposition. The results showed that the growth of the logistics industry in the four major economic regions had a positive impact on energy consumption and CO₂ emissions, with the degree of contribution being smaller in the east and central regions, and larger in the west and northeast regions. Fixed asset investment had a negative impact on energy consumption and CO₂ emissions, with the degree of contribution being largest in the northeast, larger in the east than in the central region, and smallest in the west. Finally, according to the conclusion and analysis of the results, from the aspects of government guidance and policy support, low-carbon logistics technology innovation, and infrastructure investment, we propose the need to pay attention to the role of government guidance, accelerate the pace of energy adjustments using the structure of the logistics industry, and increase the investment in renewable energy infrastructure, while focusing on strengthening the cooperation between regions and exploring new models of low-carbon logistics development between regions. This will ensure that the country achieves its goal of reaching peak carbon by 2030 and carbon neutrality by 2060.

Introduction

During the pandemic lockdowns in 2020, global carbon dioxide (CO₂) emissions fell by around 7%, or 2.4 billion tons, which was the largest absolute drop in emissions in history. While this had short-term environmental benefits, there has inevitably been a need for post-pandemic economic development, with a particular focus on how to ensure a “green recovery” of the world economy (Le et al. 2020). Due to China's economic policies and market demand, the logistics industry has developed rapidly in recent years, with a focus on optimizing industrial and economic structures, adjusting economic development models, enhancing economic development, and strengthening the interconnectivity between regions (Gunjan and Sita 2019). During 2020, China was forced to coordinate pandemic prevention and control with economic and social development. Logistics operations continued to recover steadily and total social logistics were valued at 300.1 trillion cabitale litter (46.54 trillion USD), an increase of 3.5% year-on-year at comparable prices (Xiongfeng et al. 2020; IEA 2020). While the development of China’s logistics industry is accelerating, its carbon emissions are also gradually increasing, with the logistics industry now accounting for the largest share of carbon emissions in China's total energy consumption. According to the China Energy Statistics Yearbook, carbon emissions from China's logistics industry increased to 34.61 million tons in 2017 (accounting for about 10% of China's total energy consumption and 18% of China's total carbon emissions), mainly from transportation and warehousing (Junai et al. 2019).

As an emerging complex industry, modern logistics has developed rapidly in recent years, playing an important role in guiding production and promoting consumption, but at the same time causing massive consumption of primary and secondary energy, such as coal, gasoline, fuel oil, electricity, and natural gas (Geroliminis and Carlos 2020). According to the IPCC Fifth Assessment Report, without changing the high energy consumption development of the logistics industry, by 2030 its energy consumption will increase by 80%. In December 2020, the Chinese Central Economic Work Conference stressed that China’s CO₂ emissions should strive to reach a peak by 2030 and achieve carbon neutrality by 2060. Implementing good energy-saving and emission reduction policies and accelerating low-carbon development is already one of the key elements of the Chinese government's future work. The pressure on China's logistics industry to save energy and reduce emissions has increased substantially (Xiaohong et al. 2020), and the differences between the four major economic regions of the country are significant (Junai et al. 2020). This has resulted in an urgent need to study in depth the differentiated measures among the four economic regions. Therefore, in the post-pandemic era, the establishment of dynamic and coordinated development of regional growth in the logistics industry, energy consumption, and CO₂ emissions is a key issue if China is to achieve the high-quality development of green logistics and ensure the sustainability of constructing an ecological civilization.

Studies of the relationship between the growth of the logistics industry, energy consumption, and carbon emissions have only recently been implemented. In terms of the impact of transport technologies on energy consumption, three scenarios of future transport energy use have been constructed to explore how advanced transport technologies and alternative fuels, as well as policies aimed at reducing greenhouse gas emissions and energy use, could address the energy challenges facing the transport system in the coming decades (David and Steven 2001). A positive relationship has been identified in studies of the impact of transport infrastructure renewal technologies on sustainable economic development, together with a feedback effect (Samir and Muhammad Shahbaz 2018). To determine the impact of green logistics on environmental sustainability, one previous study used panel data from 43 different countries around the world to examine the relationships between green logistics operations and energy demand, economic growth, and environmental sustainability (Syed et al. 2018). It was found that renewable energy and green development could mitigate the harmful effects of logistics operations on environmental sustainability. The impact of newly developed logistics methods on CO₂ emissions during an economic crisis was analyzed using a modified Laspeyres index decomposition (Sobrino and Monzon 2014). In terms of the dynamics of carbon emissions between the different segments of the logistics industry, the σ and β convergence methods have been applied and the highest contribution to carbon emissions was found to occur from the road transport sector (Wang et al. 2018). The impact of logistics development is believed to have played a positive role in reducing carbon emissions. Some studies have found a significant negative correlation between carbon emissions and regional logistics (Lu and Liu 2018; Sefan et al. 2018), a nonlinear variation between logistics development and CO₂ emissions (Dong et al. 2017), and a positive effect of the rapid development of the logistics industry on CO₂ emissions (Chunguang et al. 2020).

Other studies have used time series data in the form of Tapio (decoupling model), Granger (Granger) test methods, the logarithmic mean dichotomous index method (LMDI), the environmental Kuznets curve (EKC), and other methods (Samir and Muhammad Shahbaz 2018; Sobrino and Monzon 2014; Chunguang et al. 2020). Due to the uneven development of the logistics industry between different regions, the impact of growth in the logistics industry, energy consumption, and CO₂ emissions are also different; however, such panel data is not available for the different regions of China. In a regression analysis using regions as samples, the statistical results varied widely across the different economic development regions in China (Qi et al. 2020), reflecting the different levels of development in the central, east, west, and northeast economic regions (Xiaohong et al. 2016). There have been few studies of the relationship between fixed asset investment in the logistics industry and these four regions, although it is known that there is a significant positive impact of fixed asset investment on CO₂ (Wanga et al. 2019; Wangb et al. 2017).

Due to limitations such as data collection and the level of development of the logistics industries in the three different economic regions, few studies have investigated the relationships, and no unified conclusions have been reached. Therefore, this study used a panel vector autoregressive (PVAR) model based on inter-provincial panel data for the four major economic regions in China, introduced the variable of fixed asset investment in the logistics industry, and analyzed the dynamic mechanism, while controlling for individual and temporal effects (Kuan et al. 2019). The study analyzed the relationship between the growth of the logistics industry, fixed asset investment, energy consumption, and CO₂ emissions in China through Granger causality tests, an impulse response analysis, and variance decomposition results. The dynamic relationships between growth in the logistics industry, fixed asset investment, energy consumption, and CO₂ were evaluated.

Model construction

Panel vector autoregressive models have the advantage over traditional vector autoregressive (VAR) models in that all variables are set as endogenous variables, taking into account individual heterogeneity. They are now widely used in various fields of economic research (Love and Zicchino 2006). The specific analysis process is shown in Fig. 1.

Fig. 1

Panel vector autoregression (PVAR) analysis process

A PVAR model can analyze the dynamic response pathways between growth in the logistics industry, energy consumption, and CO₂ emissions (Juoudis et al. 2018; Polemis et al. 2017; Hassan et al. 2019), while compensating for the problems caused by insufficient time scales (Polemis et al. 2017). The model is shown in Eq. (1):

$$Y_{it} = B_{i} + \sum\limits_{j = 1}^{k} {C_{j} \times X_{it - j} + \mu_{i} + \tau_{t} + \varepsilon_{it} }$$

(1)

where Y_it is the column vector (each endogenous variable), X is the set of vectors of lag j order Y, B_i is the vector of the intercept terms, C_j is the coefficient matrix of the lagged variables, μ_i is the regional effect (i represents province), τ_t is the time effect (t represents year), and ε_it is the error term.

The PVAR model was used to estimate B_i and C_j. Preemptively, the within-group mean and forward mean difference methods were used to eliminate μ_i and τ_t. The validity of the estimation is more beneficial when a generalized matrix estimation was performed by applying the systematic generalized matrix estimation method (System-GMM) proposed by Abrigo (Holtz et al. 1988; Abrigo and Love 2016).

Data sources and indicator selection

Based on the availability of data, the logistics industry could be represented by data from the transport, storage, and postal industries (Lingling et al. 2019). Given that statistical data were only available from 2004 and that the data from the China Energy Statistics Yearbook is currently updated to 2018, the period 2004–2018 was selected as the research object and the study focused on the four major economic regions of east, west, central, and northeast China (National Bureau of Statistics 2020). The four major economic zones in China were delineated as shown in Table 9 in “Appendix”.

Growth in the logistics industry

According to the development status of China's logistics industry and the validity of the statistical data acquisition, this study used the value added of the logistics industry as a measure of the growth indicator of China’s logistics industry to more accurately reflect the economic growth effect of the logistics industry. Given that there is no statistical category for representing China's logistics industry (Dai and Gao 2016; Michael and Abrigo 2016), the value added of the transport, storage, and postal industry from 2004 to 2018 was used as a proxy for the value added of the China's logistics industry (Z) (Hooi et al. 2014). The value added of the logistics industry was deflated by using the GDP index with 2004 as the base period.

Energy consumption

Given that there are 16 types of energy consumption data reported for the logistics industry in the China Energy Statistics Yearbook, in accordance with the current situation and statistical data available for energy consumption in the logistics industry, eight categories were finally established (Nie et al. 2019). The physical statistics were converted into standard statistics according to the conversion coefficients given in Table 1. The energy consumption of the logistics industry in China is the sum of its converted standard coal data, which was calculated using Eq. (2):

$$E = \sum\limits_{i}^{n} {\theta_{i} \times e_{i} }$$

(2)

where θ_i is the standard coal conversion factor for the eight logistics industry energy categories, and e_i is the consumption of the eight categories of logistics industry energy.

Table 1 Standard coal conversion factors and CO₂ emission factors for various energy sources

Carbon emissions

The logistics industry in China consumes a large amount of fossil energy in transportation, distribution processing, and warehousing, which has given rise to a continuous increase in carbon emissions. To accurately measure carbon emissions from the logistics industry, the approach of Kuan et al. (2019) was in adopted to estimate carbon emissions through the indirect method. This was calculated using Eq. (3):

$$C = \sum\limits_{i = 1}^{8} {C_{i} = } \sum\limits_{i = 1}^{8} {E_{i} } \times \sigma_{i} = \sum\limits_{i = 1}^{8} {e_{i} } \times \theta_{i} \times \sigma_{i}$$

(3)

Here, the types of energy in China's logistics industry are divided into the eight categories referred to above, where C denotes the total carbon emissions of the logistics industry, e_i denotes the consumption of the ith type of energy in the logistics industry, and θ_i and σ_i denote the standard coal coefficient and carbon emission coefficient of the ith type of logistics industry energy, respectively.

Investment in fixed assets

Previous studies found that fixed asset investment had a significant impact on economic growth and energy consumption (Li and Qi 2011), promoting CO₂ emissions (Hu et al. 2012), but at the same time damaging the ecological environment. Thus, this study used fixed assets in China’s logistics industry as a control variable to measure the impact on growth in the logistics industry, energy consumption, and CO₂ exposure. Again, because there is no statistical category for the logistics industry in the existing statistical metrics, the fixed asset investment in the transport, storage, and postal industry from 2004 to 2018 was selected to replace the fixed asset investment in the logistics industry (I), and the fixed asset investment price index was used to deflate the fixed asset investment in the logistics industry (2004 as the base period).

The data were processed by taking the natural logarithm to allow the effect of heteroskedasticity to be eliminated, and the statistical indicators were the value added of the logistics industry (LnZ), fixed asset investment of the logistics industry (LnI), energy consumption of the logistics industry (LnE), and the CO₂ emissions of the logistics industry (LnC). dLnZ, dLnI, dLnE and dLnC represent the first-order differences of the above indicators respectively, and the descriptive statistics of the variables are shown in Table 2.

Table 2 Descriptive statistics

Empirical test

Panel unit root test

Given that there is some variability in the development levels of the four major economic regions in China, the common root hypothesis of the Levin–Lin–Chu (LLC), Harris–Tzavalis (HT), Breitung tests (i.e., the autoregressive coefficients are required to be equal) did not apply in practice. Therefore, the IPS, ADF-fisher, and PP-fisher tests (i.e., the autoregressive coefficients are not required to be equal) were used (Lingling et al. 2019). The Im, Pesaran and Shin (IPS), augmented Dickey–Fuller (ADF)-fisher, and Phillips–Perron (PP) unit root test results for the four major economic regions of east, west, central, and northeast China are shown in Table 3. From the statistical results, it was found that not all of the regional level data for the four major economic regions passed the IPS, ADF-fisher, and PP-fisher tests, and therefore further first-order differencing was applied, at which point all variables passed the significance level test, and the results are shown in Table 3.

Table 3 Panel unit root test

The PVAR analysis

Hysteresis order

Before constructing the model, three lag order tests, MBIC, MAIC, and MQIC, had to be applied to the PVAR model (Lanouar and Montassar 2019), and the results are shown in Table 4. To ensure the accuracy and freedom of the model, the lag order should not be too large or too small, and should be selected based on passing many tests (Gabriel et al. 2015). Therefore, order 1 was selected for all four major economic regions in China.

Table 4 Hysteresis orders

The PVAR model estimation

For the established PVAR model, this study adopted the System-GMM method, and the results are shown in Fig. 2. The results showed that the growth of the logistics industry in the eastern economic region in the previous period was significantly and positively correlated with the growth of the logistics industry, fixed asset investment, energy consumption, and CO₂ emissions in the current period, but the fixed asset investment in the logistics industry in the previous period had a suppressive effect on the growth of the logistics industry and fixed asset investment in the current period.

Fig. 2

PVAR model coefficient estimates

In the central economic region, the growth of the logistics industry in the previous period was significantly and positively correlated with the growth of the logistics industry, fixed asset investment, and energy consumption in the current period. The fixed asset investment in the previous period was significantly and negatively correlated with the growth of the logistics industry and fixed asset investment in the current period. The energy consumption in the previous period was significantly and negatively correlated with the fixed asset investment in the current period, and the CO₂ emission in the previous period was significantly and positively correlated with the fixed asset investment.

In the western economic region, the growth of the logistics industry in the previous period was significantly and positively correlated with the growth of the logistics industry, fixed asset investment, energy consumption, and CO₂ emission in the current period. The fixed asset investment in the logistics industry in the previous period was significantly and negatively correlated with fixed asset investment and CO₂ emission in the current period. The energy consumption in the previous period was significantly and negatively correlated with fixed asset investment in the current period, and the CO₂ emission in the previous period was significantly and positively correlated with fixed asset investment in the current period.

The previous period of growth in the logistics industry in the northeastern economic region was significantly and positively correlated with the current period of growth in the logistics industry, energy consumption, and CO₂ emissions. The previous period of fixed asset investment in the logistics industry was significantly and negatively correlated with the current period of growth in the logistics industry and CO₂ emissions. The previous period of energy consumption was significantly and positively correlated with the current period of fixed asset investment, and the previous period of CO₂ emissions was significantly and negatively correlated with the current period of fixed asset investment.

Granger causality test

A Granger causality test was used to analyze the causal relationship between the growth of the logistics industry, fixed asset investment, energy consumption, and CO₂ emissions. Whether there was a causal relationship between the variables was determined based on the level of significance of the coefficient estimates (a significant coefficient estimate indicated a causal relationship between the variables). The statistical results for the four major economic regions are shown in Table 5.

Table 5 Granger causality test

In the eastern economic region, the growth of the logistics industry had a two-way causal relationship with fixed asset investment, the growth of the logistics industry had a two-way causal relationship with CO₂ emissions, the growth of the logistics industry had a one-way causal relationship with energy consumption, and fixed asset investment had a one-way causal relationship with energy consumption and CO₂ emissions.

The growth of the logistics industry in the central economic region had a two-way causal relationship with fixed asset investment, the growth of the logistics industry had a one-way causal relationship with energy consumption, and fixed asset investment had a one-way causal relationship with energy consumption and CO₂ emissions.

In the western economic region, the growth of the logistics industry had a unidirectional causal relationship with fixed asset investment, energy consumption and CO₂ emissions, while fixed asset investment had a unidirectional causal relationship with energy consumption and fixed asset investment had a bidirectional causal relationship with CO₂ emissions.

The growth of the logistics industry in the northeastern economic region had a unidirectional causal relationship with fixed asset investment, energy consumption, and CO₂ emissions, a bidirectional causal relationship with fixed asset investment and CO₂ emissions, and a unidirectional causal relationship with fixed asset investment and energy consumption.

The impulse response analysis

Referring to the optimal lag order selected for China’s four major economic regions, combined with an impulse analysis, the dynamic response of each variable was investigated after the impact of a one standard deviation change (growth in the logistics industry, fixed asset investment, energy consumption, and CO₂ emissions were represented by Z1, I1, E1, and C1, respectively). The prediction period was 10 periods. The dynamic relationship between each variable was further analyzed, as shown in Figs. 3, 4, 5 and 6.

Fig. 3

Eastern regional pulse analysis

Fig. 4

Central region pulse analysis

Fig. 5

Western region pulse analysis

Fig. 6

Pulse analysis in the North East

Eastern region impulse analysis

Figure 3 shows that the pulse of CO₂ emissions on energy consumption initially resulted in a negative shock, which reached a maximum in period 1 and gradually returned to a steady state after period 2. The pulse of CO₂ emissions on fixed asset investment reached a positive peak in period 1 and then gradually returned to a steady state. The pulse of CO₂ emissions on growth in the logistics industry initially caused a positive shock, which gradually returned to a steady state after reaching a positive peak in period 1.

The pulse of energy consumption on fixed asset investment caused a negative shock, which became a positive shock after reaching a negative peak in period 1 and gradually returning to a steady state after reaching a positive peak in period 2. The pulse of energy consumption on CO₂ emissions caused a negative shock. The pulse of energy consumption on growth in the logistics industry caused a positive shock, which gradually returned to a steady state after reaching a negative peak in period 2.

The pulses of fixed asset investment in the logistics industry on CO₂ emissions and energy consumption were negative shocks, reaching a negative peak in the second period and then gradually returning to the steady state. The pulses of fixed asset investment in the logistics industry on growth in the logistics industry were negative shocks, reaching a negative peak in the first period and then gradually returning to the steady state.

The pulses of growth in the logistics industry on CO₂ emissions and energy consumption were positive shocks, reaching a positive peak in period 1 and then gradually returning to a steady state. The pulses of growth in the logistics industry on fixed asset investment were negative shocks.

Central region impulse analysis

Figure 4 shows that the pulse of CO₂ emissions on energy consumption, fixed asset investment in the logistics industry, and growth of the logistics industry initially caused a positive shock, then a negative shock after reaching a positive peak in period 1. It gradually returned to a steady state after reaching a negative peak in period 2.

The pulse of energy consumption on CO₂ emissions caused a negative shock. It gradually returned to a steady state after reaching a negative shock in period 1. The pulse of energy consumption on fixed asset investment in the logistics industry first caused a negative shock, then a positive shock after reaching a negative peak in period 1, before gradually returning to a steady state after reaching a positive peak in period 2. The pulse of energy consumption on the growth of the logistics industry caused a positive shock and gradually returned to a steady state after reaching a positive peak in period 2.

The pulses of fixed asset investment on CO₂ emissions, energy consumption, and growth in the logistics industry all caused negative shocks, reaching negative peaks in period 1 and then gradually falling back to a steady state.

The growth of the logistics industry had a negative impact on CO₂ emissions, while it had a positive impact on energy consumption and fixed asset investment in the logistics industry. It reached a positive peak in period 1 and then gradually fell back to a steady state.

Western region impulse analysis

As shown in Fig. 5, the pulse of CO₂ emissions on energy consumption initially caused a negative shock, reaching a negative peak in period 1, then rising and reaching a positive peak in period 2, before gradually falling back to a steady state. The pulse of CO₂ emissions on fixed asset investment initially caused a positive shock, reaching a positive peak in period 1 and gradually falling back to a steady state. The pulse of CO₂ emissions on growth in the logistics industry caused a negative shock, reaching a negative peak in period 1 and gradually returning to a steady state.

The pulse of energy consumption on CO₂ emissions first had a negative impact, reaching a negative peak in period 1, then rising and reaching a positive peak in period 2, before gradually falling back to a steady state. The pulse of energy consumption on fixed asset investment in the logistics industry and the growth of the logistics industry initially had a positive impact, reaching a positive peak in period 1 and then gradually returning to a steady state.

The pulse of fixed asset investment in the logistics industry on carbon emission and energy consumption caused a positive shock and gradually returned to a steady state after reaching a positive peak in period 1. The pulse of fixed asset investment on the growth of the logistics industry caused a negative shock and gradually returned to a steady state after reaching a negative peak in period 1.

The pulse of growth in the logistics industry on CO₂ emissions and energy consumption first caused a positive shock, and then gradually returned to a steady state after reaching a positive peak in period 1. The pulse of growth in the logistics industry on fixed asset investment caused a negative shock.

Northeastern region impulse analysis

Figure 6 shows that CO₂ emissions caused a positive shock on the growth of the logistics industry, gradually returning to a steady state after reaching a positive peak in period 2. The CO₂ emissions initially caused a negative shock on fixed asset investment in the logistics industry, gradually returning to a steady state after causing a negative shock in period 1.

Energy consumption had a negative impact on CO₂ emissions, reaching a negative peak in period 1 and then gradually returning to a steady state.

Fixed asset investment in the logistics industry initially had a positive impact on CO₂ emissions and energy consumption, reaching a positive peak in period 1 and then gradually returning to steady state. Fixed asset investment had a negative impact on the growth of the logistics industry, reaching a negative peak in period 1 and then gradually returning to a steady state.

The growth of the logistics industry initially had a positive impact on CO₂ emissions and energy consumption, reaching a positive peak in period 1 and then gradually returning to a steady state. The growth of the logistics industry first had a positive impact on fixed asset investment, reaching a positive peak in period 1, then falling and reaching a negative peak in period 2, before rising and reaching a positive peak in period 3 and then gradually returning to a steady state.

Variance decomposition

To measure the impact of a variable on other shocks, the degree of the main contribution of a variable to the endogenous variables was examined through a variance decomposition table (Nan et al. 2020). The forecast period was set to 20, which showed that the variance decomposition for 20 periods reached a dynamic equilibrium.

Eastern region

As can be seen from Table 6, the growth of the logistics industry was influenced by the upward trend in fixed asset investment. The initial impact was weak, but from the second period onwards the impact gradually strengthened. The later impact was significant, with a stable contribution rate of 53.52%. The growth of the logistics industry was largely influenced by energy consumption and the influence of CO₂ emissions was relatively weak.

Table 6 Results of variance decomposition in the Eastern

Fixed asset investment was gradually influenced by the growth of the logistics industry. There was a weak initial impact and a stronger impact from the second period onwards, with a stable contribution rate of 10.36%. Fixed asset investment was weakly influenced by energy consumption and CO₂ emissions.

Energy consumption was influenced by the growth of the logistics industry and the upward trend in fixed asset investment. There was a weak initial influence that gradually became stronger from the second period onwards, with the contribution rates at each stage stabilizing at 10.3% and 13.0%, respectively.

The CO₂ emissions were gradually influenced by the growth of the logistics industry and fixed asset investment, with a weak initial influence and a gradually stronger influence from the second period onwards. The contribution rates at each stage stabilized at 9.7% and 12.0%, respectively. Initially, CO₂ emissions were mainly influenced by energy consumption, with a gradually weaker influence from the second period onwards, and the contribution rate stabilizing at 75.2%.

Central region

As can be seen from Fig. 7 the growth of the logistics industry was influenced by the upward trend in fixed asset investment, with a weak initial influence and a gradually stronger influence from the second period onwards. There was a significant influence at the later stage and a stable contribution rate of 47.3%. The growth of the logistics industry was weakly influenced by energy consumption and CO₂ emissions.

Fig. 7

Results of variance decomposition in the central

Fixed asset investment was gradually influenced by the growth of the logistics industry, with a weak initial impact and a gradually stronger impact from the second period onwards. There was a stable contribution rate of 15.0%. Fixed asset investment was weakly influenced by energy consumption and CO₂ emissions.

Energy consumption was influenced by the growth of the logistics industry and the upward trend in fixed asset investment, with a weak initial influence and a gradually stronger influence from the second period onwards. The contribution rates at each stage stabilized at 6.2% and 7.1%, respectively.

The CO₂ emissions were gradually influenced by the growth of the logistics industry and fixed asset investment, with a weak initial influence and a gradually stronger influence from the second period onwards. The contribution rates at each stage stabilized at 5.7% and 5.2%, respectively. The CO₂ emissions were mainly influenced by energy consumption, and gradually weakened from the second period onwards, with the contribution rate stabilizing at 85.7%.

Western region

As can be seen from Table 7 the growth of the logistics industry was mainly influenced by itself and had little relationship with other factors.

Table 7 Results of variance decomposition in the Western (stable contribution after period 2)

Fixed asset investment was gradually influenced by the growth of the logistics industry itself, with a weak initial influence. From the second period onwards the influence gradually strengthened and the contribution rate was stable at 26.0%. Fixed asset investment was only weakly influenced by energy consumption and CO₂ emissions.

Energy consumption was influenced by the growth of the logistics industry and the upward trend in fixed asset investment, with a weak initial influence and a gradually stronger influence from the second period onwards. The contribution rates in the two stages stabilized at 15.1% and 3.0%, respectively.

The CO₂ emissions were gradually influenced by the growth of the logistics industry and fixed asset investments, with a weak initial influence and a gradually stronger influence from the second period onwards. The contribution rate in the two stages stabilized at 15.7% and 2.9%, respectively. Initially, CO₂ emissions were mainly influenced by energy consumption, with a gradually weaker influence from the second period onwards. The contribution rate stabilized at 78.7%.

Northeastern region

As can be seen from Table 8 the growth of the logistics industry was influenced by the upward trend in fixed asset investment, with a weak initial influence and a gradually stronger influence from the second period onwards. There was a significant influence in the later period, with a stable contribution rate of 38.5%. The growth of the logistics industry was only weakly influenced by energy consumption and CO₂ emissions.

Table 8 Results of variance decomposition in the North East (stable contribution after period 2)

Fixed asset investment was gradually influenced by the growth of the logistics industry and CO₂ emissions, with a weak initial impact and a gradually stronger impact from the second period onwards. The contribution rates in the two stages stabilized at 7.5% and 10.6%, respectively.

Energy consumption in the current period was mainly influenced by itself, the growth of the logistics industry, and fixed asset investment. From the second period onwards, the influence of the growth of the logistics industry and fixed asset investment increased. The contribution rates in the two stages stabilized at 24.4% and 42.1%, respectively.

The CO₂ emissions in the initial period were mainly influenced by the growth of the logistics industry, fixed asset investment, and energy consumption. From the second period onwards, the influence of the growth of the logistics industry, and fixed asset investment gradually strengthened. The contribution rates in the two stages stabilized at 23.5% and 44.3%, respectively. The contribution from energy consumption gradually weakened after the second period, with a stable rate of 26.9%.

Conclusions and recommendations

Conclusions

Based on inter-provincial panel data for China’s logistics industry from 2004 to 2018, a PVAR model was developed to test the dynamic transmission mechanisms of growth in the logistics industry, energy consumption, and carbon emissions in the eastern, central, western, and northeastern regions. The main findings were as follows.

• 1. From the perspective of the eastern region, there was a one-way causal relationship between the growth of China's logistics industry and energy consumption, and there was a two-way positive impact of the two factors. There was a two-way causal relationship between the growth of the logistics industry and CO₂ emissions, and there was a two-way positive impact of the two factors. Fixed asset investment had a negative impact on the growth of the logistics industry, energy consumption, and CO₂ emissions.

• 2. From the perspective of the central region, there was a one-way causal relationship between the growth of China’s logistics industry and energy consumption, and there was a two-way positive impact of the two factors, although it was weaker than that in the east. Fixed asset investment had a negative impact on energy consumption and CO₂ emissions, although again it was weaker than that in the east.

• 3. From the perspective of the western region, there was a one-way causal relationship between the growth of China's logistics industry and energy consumption. There was a positive impact of the two factors, and the degree of impact was stronger than that in the eastern and central regions. The CO₂ emissions were affected by the positive impact of the growth of the logistics industry, and the degree of impact was stronger than that in the eastern and central regions. Fixed asset investment had a negative impact on energy consumption and CO₂ emissions, and the degree of impact was weaker than that in the eastern and central regions.

• 4. From the perspective of northeast China, there was a one-way causal relationship between the growth of China’s logistics industry and energy consumption, and there was a positive impact of the two factors, which was stronger than that in the eastern, central, and western regions. There was a one-way causal relationship between the growth of the logistics industry and CO₂ emissions. There was a positive impact of the two factors, and the degree of impact was stronger than that in the eastern, central, and western regions. The negative impact of fixed asset investment on CO₂ emissions and energy consumption was significantly stronger than that in the eastern, central, and western regions.

Recommendations

Based on the uneven development between the different economic regions, to resolve the conflict between economic growth and CO₂ emissions in the logistics industry, China should actively explore interregional low-carbon logistics development models to achieve high-quality sustainable development. In this regard, the following recommendations were made.

• 1. From the perspective of the eastern region, fixed asset investment will not promote the growth of the logistics industry in the eastern region, but it could effectively inhibit energy consumption and CO₂ emissions. In recent years, the logistics industry in the eastern region has developed rapidly, actively responding to the call for “low-carbon logistics with energy conservation, and emission reduction, green development, and environmental protection”, and reducing the impact on the environment (Guanxin et al. 2018). While ensuring the growth of the logistics industry, we should continue to actively increase the investment and construction of new technologies, new equipment, and other infrastructure, further optimize the energy structure and industrial structure, and vigorously develop low-carbon logistics.

• 2. From the perspective of the central region, although fixed asset investment will not promote the growth of the logistics industry, it could inhibit CO₂ emissions. The logistics industry in the central region is growing rapidly due to the obvious regional advantages. At the same time, there is a convenient transportation network and the development of low-carbon logistics is beginning to show results. While vigorously developing the logistics industry, we should further combine the regional advantages, rely on the development strategy of the Yangtze River economic belt, improve the level of science and technology education, and actively promote technical energy conservation and emission reduction measures in transportation, warehousing, and other aspects of the logistics industry (Jiang et al. 2020).

• 3. From the perspective of the western region, the growth of the logistics industry will boost CO₂ emissions, and the fixed asset investment of the logistics industry will not effectively curb regional carbon emissions. The western region should therefore focus on vigorously developing the logistics industry and improving the level of economic activity. At the same time, there is a need to learn from the low-carbon logistics emission reduction technology and modern logistics development mode in the eastern region to speed up the transformation and upgrading of the industry to ensure green logistics. While paying attention to the growth of logistics industry, combined with the unique regional advantages of the western region, it should be possible to attract more talented professionals and scientific research investment, improve the level of scientific and technological innovation, and promote the large-scale operation of the logistics industry (Zhenggang et al. 2017).

• 4. Compared with other economic regions, the growth of the logistics industry in the northeast economic region contributed the most to energy consumption and CO₂ emissions. Fixed asset investment could effectively curb petrochemical energy consumption and CO₂ emissions. In the context of the strategic revitalization of the northeast, the northeast economic region needs to accelerate the adjustment of industrial structure, promote industrial optimization and upgrading (Fan and Hao 2020), change the development mode of high energy consumption and high emissions, and eliminate poorly developed production practices (Bingquan et al. 2020; Zuankuo et al. 2019). There should be a focus on the development of a modern logistics intelligent infrastructure network in the context of “Internet + ”, vigorous development of a modern logistics industry, and the gradual formation a low-carbon logistics system.

Appendix

See Table 9.

Table 9 The division of China's four major economic zones