Measurement of the Resilience of China’s Logistics Industry and Its Influencing Factors

Abstract

Improving the resilience of the logistics industry to enhance its ability to withstand a variety of uncertain risks is critical to its development. However, the spatio-temporal evolution of resilience in the logistics industry has not been adequately characterized in the literature. And the factors influencing it remain unclear. This study analyzes the patterns of spatio-temporal evolution of the resilience of the logistics industry in China’s 31 provinces from 2012 to 2021 to identify the factors influencing it. This research found that the overall resilience of the logistics industry has been increasing with a pattern in which the east regions are high, the west regions are low, the south regions are high, and the north regions are low, while the coastal levels are generally higher. However, the levels of resilience and adaptability of the logistics industry are both low and fragmented and need to be improved over the short term. This study emphasizes promoting projects to improve the logistics infrastructure and increasing fixed-asset investment in the logistics industry to improve the resilience of the logistics industry in China.

1. Introduction

The logistics industry is a prominent part of the national economy and a vital driver of economic growth. In recent years, disasters have had a profound global impact. Sudden major public crises, such as the recent COVID-19 pandemic, have had a significant impact on all walks of life. As a fundamental sector of the national economy, the logistics industry is no exception to disruptions to supply chains and rising freight costs on a global scale. China’s total imports accounted for CNY 17.4 trillion in 2021, a decline of 1% compared with the previous year, but increased by 13% in 2023 [1]. Improving the resilience of the logistics industry is necessary for its coordinated development with the urban economy [2]. Measuring its resilience is therefore important for the development of the logistics industry in order to reduce its vulnerability to geopolitics, natural disasters, and ensuing uncertainties.

Most research on resilience has focused on evaluations of municipal economic resilience [3] and industrial resilience [4], primarily by using quantitative evaluations of the level of resilience based on entropy weighting [5], the TOPSIS model [6], and the geographic detector method [7]. For example, Bruneckiene (2018) established a system of indicators to measure resilience along six dimensions including governance, innovation, learning, cooperation, infrastructure, and regional insight [8]. Wang (2023) evaluated the durability of the industrial economy of the Taihu Lake Basin (TLB) during the global financial crisis of 2008 [9]. They assessed it by a multi-indicator, comprehensive evaluation system that made use of spatial analysis based on geographic information systems. They used sensitivity to resistance as the major variable and applied the entropy assignment technique to evaluate the financial durability of the region fully. Ye (2023) analyzed the impact of the structure of industrial linkages on the financial durability of China’s cities under the shock of the COVID-19 pandemic [10]. Cai and Xu (2022) used the fixed-base coefficient of efficacy to measure the composite index of the industrial resilience of Chinese provinces in terms of their dynamic evolution and discretized spatial distribution from 2007 to 2019 [11]. Jiang and Jing (2022) used the shift-share method of decomposition to reveal the mechanism of the competitiveness of the structure of the industry in terms of its economic resilience. They used the geodetector model to identify the drivers of the latter [12]. Wang et al. (2022) explored the coupled association between financial agglomeration and economic resilience by using cluster analysis and the entropy weighting method [13]. Researchers have thus used quantitative measures to focus on the spatio-temporal patterns of the evolution of resilience as well as the factors influencing it.

Scholars have found that the same factors, including the capabilities for innovation, have different influences on resilience in different industries. For example, Li et al. (2024) used panel data on 297 cities in China from 2015 to 2020 to construct fixed-effects and spatial econometric models. They concluded that the servicification of manufacturing had a positive direct intra-regional effect and a negative, indirect extra-regional effect on the economic resilience of cities [14]. Peng et al. (2023) constructed a system to measure agricultural economic resilience to explore its impact, as well as the impact of factors influencing it, on agricultural development. They concluded that improving scientific and technological innovation is conducive to agricultural development [15]. Huang (2023) constructed a system of indices to assess 282 Chinese cities’ economic resiliencies between 2005 and 2019. They found that differences in the capabilities for technological innovation were the key causes of the geographic differences in economic tenacity [16].

Individual elements have distinct effects when they are present in the same sector. Tan et al. (2023) used econometric models, including binary logit, to analyze the factors influencing industrial resilience from the perspectives of the industrial structure and regional environment. They concluded that specialization had a negative impact on the resilience of labor-intensive industries, while unrelated diversification facilitated it [17]. Jiang et al. (2023) found that changes in the price of agricultural products had a significant and gradually increasing influence on the resilience of food production [18]. Yu et al. (2021) examined the spatio-temporal heterogeneity in the timelines of logistics from a resilience perspective. They identified a pronouncedly unfavorable relationship between the centrality of the nodes of the logistics network and the composite index of resilience. Moreover, they concluded that the effect of the correlation between the level of development of the logistics industry and the strength of epidemic prevention and control in the context of its resilience was not clear [19]. Ye et al. (2024) investigated the factors influencing the level of high-quality and green economic development in ports. They found that factors such as the level of scientific and technological progress caused a notable improvement in them, while the industrial structure had a significant negative effect [20]. It is thus evident that different factors in different industries have varying effects on the resilience of the logistics network and thus need to be measured separately.

Because of the importance of the logistics industry for economic development, its resilience has attracted considerable attention in research, which has largely focused on quantitative analyses. Researchers have mainly used its capabilities of resistance to risk, adaptivity and restoration, and innovation and upgrade to measure the resilience of the logistics sector. Ma, Hou, and Yang (2022) used these measures to construct a system of indices to assess the resistance of the logistics industry in 29 provinces of China from 2010 to 2020 [21]. They used focus interviews, the Delphi method, and multi-criteria decision-making for measurement. Moreover, Gupta (2022) argued that the role of logistics has become more important in a wider context in recent years. They analyzed five major barriers to the enforcement of innovative digital technologies in this context for developing countries by using multi-criteria decision analysis methods [22]. Sun and Chawalit (2021) conducted semi-structured interviews with the top executives of three companies. They identified the key factors supporting the resilience of logistics firms during the outbreak in terms of their flexibility, business continuity plans, and diversification of markets [23].

In summary, owing to different research-related perspectives on urban economic resilience and the resilience of different industries, researchers have not yet arrived at a uniform system of indicators and dimensions for measuring resilience. From the standpoint of its ability to restructure the logistics industry, the ability of the sector to resist, adapt, recover, and renew is more indicative of its short-term capability to confront shocks. Therefore, it is important to develop a system of indices to assess the resilience of the logistics industry along the dimensions of resistance, adaptation, recovery, and capacity and further renewal to inform research in the area. Moreover, the spatio-temporal evolution of the resilience of the logistics industry needs to be further examined. Empirical research in the area has mostly focused on its economic and urban resilience. In addition, recent research on the resilience of the logistics industry has primarily considered the choice of methods of measurement and the identification of influential factors without integrating resilience with spatial theoretical methods. This makes it difficult to reflect the spatio-temporal differences in the resilience of the logistics industry comprehensively. Therefore, identifying the factors influencing the level of resilience of the logistics industry is crucial for its development.

The primary contributions of this paper include the following: (1) We establish a system of indices to assess the resilience of the logistics sector using a systematic analysis based on resilience theory. We then use it to quantify the capacity of the logistics industry for resilience in 31 provincial-level administrative regions in China. (2) We combine ArcGIS 10.6 software, the spatial autocorrelation model, and the kernel density estimation model to explore its spatio-temporal evolution. (3) We use spatial econometric modeling to identify the factors that enhance the resilience of the logistics sector. This study provides guidance for the logistics industry in China and other countries to achieve sustainable development. It also provides a model-based reference for research on enhancing the resilience of other industries.

The remainder of this paper is structured as follows: Section 2 introduces the methodology and data used to measure the resilience of China’s logistics industry. Section 3 presents the results of measurements of the dimensions of resilience based on spatio-temporal evolutionary analyses as well as spatial autocorrelation tests. It also details regression analyses of selected fixed-effects models to identify the factors influencing their outcomes. Section 4 discusses past work in the area, and Section 5 summarizes the conclusions of this study.

2. Materials and Data

2.1. Entropy

Single and composite indicator frameworks are currently among the most used resilience frameworks. The assessment of logistics resilience using a single indicator is simpler to use but falls short of accurately capturing the complexity and openness of the system. Integrated evaluation is a broader approach to evaluating complex systems using a number of indicators. In order to describe logistics resilience more comprehensively, this paper uses multiple indicators and chooses a more comprehensive indicator system to assess logistics resilience. Various methods are used to calculate weights in comprehensive multi-indicator assessments, and the differences among them mainly lie in the focus of the assignment of indicators. Principal component analysis, data envelopment analysis, factor analysis, hierarchical analysis, the entropy value method, etc., are mostly used in the current research.

Principal component analysis and factor analysis are used to determine the weights of the indicators based on empirical data so that they are objective and can reduce the dimensionality of the data. However, if the sign of the load of the main component is positive or negative, its meaning is unclear. If the analysis of the main components of the system of indicators leads to a situation where the factor loadings are negative, the relevant method of weighting is excluded from consideration. The entropy value method is a scientific technique that can avoid unknown errors and the influence of human factors. It can objectively represent the importance of indicators in the system of assessment and reflect the resulting changes in their weights [24]. Through this approach, more accurate results can be effectively obtained, reflecting more realistically the importance of the indicators in the indicator framework and the corresponding changes in weighting. The entropy value method can not only avoid the above problems but can also reasonably evaluate the resilience of the logistics industry. Therefore, the entropy method was used in this paper to provide a comprehensive evaluation of the resilience level of China’s logistics industry.

Through the entropy method, the weights of the indicators at each level were determined. The calculation steps are as follows:

The data are standardized in the first step as follows:

$${\mathrm{x}\prime }_{ij}=\frac{x_{ij}-\min (x_{ij})}{\max (x_{ij})-\min (x_{ij})}(\mathrm{positive} \mathrm{indicator})$$

(1)

$${\mathrm{x}\prime }_{ij}=\frac{\max (x_{ij})-x_{ij}}{\max (x_{ij})-\min (x_{ij})}(\mathrm{negative} \mathrm{indicator})$$

(2)

where i = 1, 2, …, n; j = 1, 2, …, m. x_ij is the value of the jth indicator in year i and is normalized in the range of [0, 1]. max(x_ij) is the maximum value of x_ij, and min(x_ij) is its minimum value.

The second step involves finding the proportion of each indicator. The calculation is as follows:

$$P_{ij}=\frac{{x\prime }_{ij}}{{\displaystyle {\sum }_{i=1}^n{x\prime }_{ij}}}$$

(3)

The third step involves finding the informational entropy of each indicator:

$$E_j=\frac{1}{\ln n}{\displaystyle {\sum }_{i=1}^n{P}_{ij}\ln P_{ij}}$$

(4)

The weights are calculated by calculating the redundancy of information:

$$D_j=1-E_j$$

(5)

The indicator weights are calculated as:

$$W_j=\frac{D_j}{{\displaystyle {\sum }_{j=1}^m{D}_j}}$$

(6)

The single-indicator evaluation scores are calculated as follows:

$$S_{ij}=W_j\times X_{ij}$$

(7)

where m is the number of indicators, Sij is the proportion of the jth index in year i, and Ej, Wj, and Dj are the entropy, weight, and coefficient of variation of the jth index, respectively.

2.2. Model to Analyze the Spatio-Temporal Level of Resilience of the Logistics Industry

2.2.1. Spatial Autocorrelation Model

Global spatial autocorrelation analysis describes the spatial aggregation or distribution of a given area with respect to the importance of evaluation. The global Moran’s I is commonly used to reflect the degree of development of regional spatial correlation [25]. This method was used in this paper to analyze the spatial correlation of China’s logistics industry. Moran’s I index takes the range [−1, 1], where −1 < Moran’s I < 0 means that values of the evaluated attributes of the given object are spatially negatively correlated. Moran’s I=0 means that they are almost uncorrelated, and 0 < Moran’s I < 1 means that the values of attributes of the evaluated object are spatially positively correlated:

$$I=\frac{n}{S_0}\cdot \frac{{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{w}_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}}}{{\displaystyle \sum _j^n{\left(x_i-\overline{x}\right)}^2}}$$

(8)

where Xi denotes the observation at the first spatial location, $\overline{x}=\frac{1}{n}{\displaystyle \sum _{i=1}^n{x}_i}\underset{x\to \infty }{\lim }$, W_ij is an element of spatial weights that represent topological interactions among the spatial units, n is the total number of elements, which are all equal, and S₀ is the sum of elements of the spatial inter-weight matrix W.

Global autocorrelation analysis describes the spatial distribution of the object of interest but cannot describe the characteristics of clustering of the local space within the field of research. Local autocorrelation is used to describe this instead by using the local Moran’s I [26]. Its value and properties are the same as those of the global Moran’s I.

Moran’s scatterplot represents the observed values of the variable at each place on the horizontal axis and the spatial lag on the vertical axis. It is used to measure local spatial autocorrelation, whereas Moran’s I is used to quantify the global spatial autocorrelation of regional economic behavior. The correlation between them is represented by the scatterplots in the coordinate system [27]. The scatterplots may be distributed in any of the four quadrants of the coordinate system, and each quadrant symbolizes a different sense. The first (H-H) and third quadrants (L-L) correspond to a positive spatial autocorrelation, indicating that the observations at the given location are similar to those at its neighbors. A negative spatial autocorrelation is correlated with the second (L-H) and fourth (H-L) quadrants, suggesting that observations at the specified position are not similar to those at their neighbors.

2.2.2. Kernel Density Estimation Model

Kernel density estimation treats the spatial distribution of the given object as a probability distribution to reveal the evolution of the morphological characteristics of its distribution over time [28]. Therefore, the kernel density estimation model was used in this paper to analyze the distributional dynamics and polarization phenomena of the coupled coordination among urban resilience subsystems in 2012–2021.

The Gaussian kernel function in the kernel density estimation model was selected for analysis in this paper and is calculated as follows:

$$f\left(xi\right)=\frac{1}{nh}{\displaystyle {\sum }_{i=1}^{n}K}\{\frac{xit-\overline{x}}{h}\}$$

(9)

where f(xi) is the density function of a random variable x in year t, n is the number of regions, h is the bandwidth, and K(*) is the kernel density function.

2.3. Spatial Measurement Models

Spatial econometric modeling was used in this paper to measure the factors influencing the level of resilience in the logistics industry, including the spatial lag model (SAR), the spatial Durbin model (SDM), and the spatial error model (SEM). It is a data model and method for empirical analysis of the spatial distribution of socio-economic types using spatial features and spatial correlations [29]. It not only helps us comprehend how space, society, and the economy are related, but it also makes insightful recommendations for resolving geographic issues. Spatial econometric modeling is an important tool in the analysis of and research on geospatial data. The spatial lag model assumes that the values of the variables in one region are influenced by those in its neighboring regions, i.e., there are interactions among spatially neighboring areas. This is consistent with China’s current situation. Therefore, the spatial lag model was used in this paper to calculate the factors influencing the level of resilience of the logistics industry.

The SAR is mainly applicable when there is a strong and direct spatial dependence of the explanatory variables. This is due to the fact that neighboring explanatory variables influence local explanatory variables through the mechanism of spatial transmission. This is formulated as:

$$y=\lambda Wy+X\beta +{\epsilon }_y$$

(10)

where y is the dependent variable and X is the independent variable. W represents the matrix of spatial weights. λ is the coefficient of spatial weights. β is the regression coefficient of the independent variable X. ε is the error term. The matrix of spatial weights is then given as follows:

$$W_{ij}=\left[\begin{array}{ccc}{w}_{11}& \cdots & w_{1n}\\ ⋮& \ddots & ⋮\\ w_{n1}& \cdots & w_{nn}\end{array}\right]$$

(11)

When i is equal to j, W_ij = 0.

The elements of the matrix after normalization are as follows:

$$W_{ij}=w_{ij}/{\displaystyle {\sum }_{j=1}^n{w}_{ij}}$$

(12)

2.4. Variables and Data

2.4.1. Variable Selection and Weighting for Measuring the Level of Resilience of the Logistics Industry

We referred to the relevant literature and drew on systems of indices for the assessment of resilience according to resistance, adaptability, and resilience dimensions proposed by Jiang and Wang [30,31]. We combined it with the current stringent requirements of the logistics industry for operational efficiency, infrastructure for further improvement, and a high service level, as well as its features of slow innovation and development. We created a system of indices for evaluation along four dimensions including the resistance, adaptability, resilience, and capacity of update of the logistics industry. We derived the weights of 24 Tier 1 indicators and four dimensions of Tier 1 indicators. They are shown in

Table 1. Weights of indices in the system to evaluate the resilience of the logistics industry.

Level-1 Indicators	Secondary Indicators	Resilience Evaluation Index	Weight
Resistance	Production	Per capita GDP (CNY)	0.013244
	Logistics economy	Investment in fixed assets in the logistics industry (CNY 100 million)	0.025896
		Operating income from the logistics industry (CNY 100 million)	0.033157
Adaptive capacity	Logistics infrastructure	Miles of road for traffic (km)	0.01819
		Railway mileage (km)	0.025052
		Inland waterway mileage (km)	0.057806
		Civil aviation route mileage (km)	0.057636
	Logistics demand	Volume of cargo transported (10,000 tons)	0.021458
		Cargo turnover (100 million ton-km)	0.036314
		Total post and telecommunications operations (CNY 100 million)	0.076724
	Information technology level	Mobile phone penetration rate (department/100 people)	0.045874
		Internet penetration rate (%)	0.040133
	Labor level	Number of employees in the logistics industry (10,000 people)	0.023137
		Full-time equivalent of logistics R&D personnel (person-year)	0.03508
		Quality of labor in the logistics industry	0.025263
Recovery capability	Logistics scale	Value added of the logistics industry (CNY 100 million)	0.084006
	Operational efficiency	On-time delivery ratio (%)	0.03613
		Delivery breakage ratio (%)	0.024926
		Inventory turnover ratio (%)	0.051334
		Order fulfilment rate (%)	0.040283
		After-sales service efficiency (%)	0.032676
Updating capacity	Capital investment	Scientific and technological inputs to the logistics industry (CNY 100 million)	0.062815
		Logistics science and technology expenditure (CNY 100 million)	0.053703
	Technological achievements	Number of patents granted in logistics (CNY 10,000)	0.056132

.

2.4.2. Factors Influencing the Level of Resilience of the Logistics Industry

Our selection of the factors influencing the resilience of the logistics industry was based on the relevant internal and external environments. The external environment mainly considers the level of economic development, while internal environmental factors include the technological level. The selection was based on a review of the literature on the factors influencing the resilience of the logistics industry, its current status, and the reliability of the indicator-related data. We analyzed the following six dimensions: economic foundation, government support, capacity for innovation, openness to the world, environmental pollution, and infrastructure [32]. They are shown in

Table 2. Selection and description of the influential factors and indicators.

Influential Factors	SN	Variables	Units
Economic foundation	X1	Per capita GDP	CNY
Government support	X2	R&D expenses	CNY 10,000
Innovation ability	X3	Number of valid invention patents	Pieces
Openness to world	X4	Total investment	USD billions
Environmental pollution	X5	Waste gas pollutant emissions	10,000 tons
Infrastructure development	X6	Total length of postal routes	km

.

2.4.3. Data Sources

We examined the logistics industry in 31 provinces of China, excluding Hong Kong, Macao, and Taiwan, from 2012 to 2021. Panel data for all provinces from 2012 to 2021 was used to this end. All data were obtained from the China Logistics Yearbook, the National Statistical Yearbook, the National Bureau of Statistics, and provincial statistical yearbooks and statistical bulletins. The neighboring year value or assignment method was used to fill in missing data for individual provinces.

3. Results

3.1. Resilience of China’s Logistics Industry and Its Spatio-Temporal Evolution

3.1.1. Dimensional Resilience

(1)

Level of resistance

The results of the evaluation of the levels of resistance-related resilience in the 31 provinces of China in 2012–2021 are shown in Appendix A. We calculated the average resistance-related resilience of each province in the study period as well as the overall national average, and the results are shown in Figure 1 and Figure 2. It is evident that Guangdong obtained the highest resistance score, while Tibet had the lowest. The mean value of resistance in 10 provinces, including Guangdong, Shandong, Zhejiang, Jiangsu, Hubei, and Sichuan, was above 0.4, which represents a high level of resistance. Eighteen provinces had mean values in the range of 0.1–0.4, which indicates that their resistance-related resilience needed to be improved. The mean resistance of three other provinces was below 0.1, suggesting that they had insufficient resistance. On the whole, the level of the capacity for resistance grew rapidly across China in 2012–2018 and leveled off in 2018–2021.

(2)

Adaptive capacity

The results of the evaluation of the adaptive capacity of the 31 provinces of China in 2012–2021 are shown in Appendix A. We calculated the average adaptive capacity of each province as well as the national average in the study period, as shown in Figure 3 and Figure 4. Provinces with an adaptive capacity higher than 0.4, which represents a strong adaptive capacity, included Hebei, Liaoning, Jiangsu, Zhejiang, Shandong, Henan, Hubei, Guangdong, and Sichuan. A total of 21 provinces had mean values of adaptive capacity ranging from 0.1 to 0.4, indicating that their adaptive capacity needed to be improved. The adaptive capacity of Tibet was below 0.1, indicating that it needed to improve its level of information technology to improve its adaptive capacity. On the whole, the level of adaptive capacity across China fluctuated considerably, with minor differences among provinces.

(3)

Capacity for recovery

The results of the evaluation of the capacity for recovery of the logistics sector in China’s 31 provinces in 2012–2021 are shown in Appendix A. We calculated the average capacity of the 31 provinces for recovery in 2012–2021 as well as the national average, and the outcomes are displayed in Figure 5 and Figure 6. It is evident that the logistics industry in 10 provinces, including Hebei, Shanghai, Jiangsu, Shandong, and Guangdong, had a high capacity for recovery (>0.4). Jilin, Hainan, Chongqing, Guizhou, and most other provinces had a low capacity for recovery, as indicated by their scores falling in the range of 0.1–0.4. The gap among provinces in this regard was large, and their logistics industries needed to increase coordination with one another to restore operational efficiency. Tibet had the lowest level of capacity for recovery at 0.045. Overall, the capacity of the logistics sector for recovery in China’s 31 provinces exhibited slow growth in the study period, with a relatively high and steadily increasing level of resilience.

(4)

Capacity for update

The results of the evaluation of the capacity of the logistics sector for renewal in China’s 31 provinces in 2012–2021 are shown in Appendix A. We calculated the average values of the capacity for renewal for all 31 provinces as well as the national average, and the results are shown in Figure 7 and Figure 8. They show that the value of the capacity of the logistics industry for renewal was above 0.4 in Zhejiang and Guangdong. It was in the range of 0.1–0.4 in Beijing, Fujian, Henan, Hubei, and nine other provinces. This suggests that these regions should attend to scientific and technological innovation to improve their capacity for updating. The regions of Tibet, Hainan, Qinghai, and Ningxia received the lowest marks, indicating that these provinces had low levels of renewability-related resilience and needed to improve their use of resources. The capacity for updating the logistics sector in China’s 31 provinces exhibited a significant increase over the study period. Overall, the level of capacity for renewal was low, even though it was increasing.

3.1.2. Comprehensive Results

The results of the evaluation of the comprehensive level of resilience of the logistics sector in China’s 31 provinces in 2012–2021 are shown in Appendix A. We calculated the average scores for all 31 provinces in 2012–2021 as well as the national average, as shown in Figure 9 and Figure 10. It is evident that the level of resilience of the logistics industry among provinces was unbalanced, with Guangdong obtaining the highest score of 0.69 and Tibet with the lowest of 0.02. Moreover, the resilience of the logistics industries of the provinces grew at different rates from 2012 to 2021, and its annual average amount increased from 0.20 in 2012 to 0.28 in 2021. This suggests that the degree of resilience of the national logistics sector increased and is sound on the whole. Under the backdrop of rapid economic development, the logistics industry in all provinces improved its ability to withstand risks and disasters as well as its ability to recover.

3.1.3. Spatio-Temporal Evolution of Subdimensions of Resilience

ArcGIS software was used in this paper to make a map of the distribution of the resilience level of the logistics industry in 31 provinces in China from 2012 to 2021. Accordingly, we further explored the characteristics of the spatial evolution pattern of the resilience level of each dimension of China’s logistics industry. We defined four data points based on the measured data from each province and categorized the resilience of the logistics industry into four levels. They are as follows: high-level area, higher-level area, lower-level area, and low-level area—according to the natural breakpoint approach. We selected four time points—2012, 2015, 2018, and 2021—and plotted the spatial distribution visualization of each dimension. This shows the spatio-temporal evolution of the resilience level of China’s logistics industry. The outcomes are displayed in Figure 11, Figure 12, Figure 13 and Figure 14.

The spatial distribution of the resilience level of the logistics industry in the years 2012, 2015, 2018, and 2021 is shown in Figure 11. The trend of growth in its levels of resilience in Yunnan, Sichuan, Jiangsu, Fujian, and Zhejiang provinces was prominent from 2012 to 2021. However, the spatial distribution of high-value areas of resilience increased. The level of resilience of each province’s logistics sector was in the low range of values in 2012. Only Jiangsu, Shanghai, Zhejiang, Guangdong, Shandong, Hebei, and Beijing had a higher degree of tenacity. In 2021, the tenacity of the logistics industry exhibited clear regional differentiation. It was weaker in the northwestern and northeastern regions of each province and municipality, where Beijing, Shandong, and Guangdong provinces had high resilience. On the whole, the resistance-related resilience of the logistics industry changed significantly in 2012–2021, and it exhibited a trend of agglomeration to the southeast. The resilience of the southeastern provinces was generally higher, while that of the western regions, including Tibet, Qinghai, and Gansu, was lower. This suggests a lack of investment in infrastructure in the western region. This needs to be addressed in future policymaking on the issue.

The spatial distribution of the level of adaptive resilience of the logistics industry for the four years 2012, 2015, 2018, and 2021 is shown in Figure 12. The geographical arrangement of zones with high values of adaptive capacity for the logistics industry gradually increased from 2012 to 2021, with a gradual divergence in 2021. In 2012, the adaptive capacity of the logistics industry had low values, while high-value areas were primarily distributed in the central part of the country. Compared with that in 2012, the logistics sector’s ability to adapt in 2015 was more prominent in the southwestern part of the country but did not improve by much in its southeastern part. In 2021, the adaptive capacity of the logistics industry significantly changed in Anhui, Hunan, Shanxi, and particularly in Guizhou province. However, Xinjiang, Chongqing, Jiangxi, Jiangsu, and Zhejiang provinces exhibited no significant change. Overall, the adaptive resilience of the logistics sector in the east of China had a strong foundation and gradually improved.

The spatial distribution of resilience levels of the logistics industry for the four years 2012, 2015, 2018, and 2021 is shown in Figure 13. From 2012 to 2021, its levels of resilience in Xinjiang, Jiangxi, Guizhou, Guangxi, and other provinces grew significantly. The quantity of zones having a substantial degree of resilience grew year by year and expanded to the country’s western and central regions. In 2012, the resilience of the logistics industry was in the low range. But the overall level of the eastern provinces’ resilience was higher and landed in an area of considerable value. Relative to 2012, the enhancement in the levels of resilience of the logistics industry in Jiangsu, Hunan, Jiangxi, and other provinces in 2015 was clear. The central and eastern parts gradually assumed higher values. The degree of resilience of the logistics sector in most provinces was high in 2021. But Tibet, Qinghai, Gansu, Ningxia, and certain other provinces still had low values, demonstrating the logistics sector’s resilience in these areas needed to be improved.

The spatial distributions of the capacity of the logistics industry for renewal in 2012, 2015, 2018, and 2021 are shown in Figure 14. It shows that from 2012 to 2021, the geographic dispersion of regions where the logistics sector was present had a high capacity for renewal, which decreased while the quantity of places with high capabilities in this respect rose. The distribution of such areas mainly increased in the center and the southeast, while the number and patterns in the distribution of areas with a low capacity for renewal remained unchanged. The resilience of the logistics industry’s capacity for renewal in 2012 was in the range of low values, except in Sichuan, Shandong, Jiangsu, Zhejiang, and Guangdong, which had high values. Relative to that in 2012, there was no significant change in the logistics industry’s capacity for renewal in 2015. The regional differentiation of the renewal capacity toughness level of the logistics industry in 2021 was obvious, and the toughness level of the central and coastal regions was generally high. The northwest, southwest, and northeast regions, such as Tibet, Qinghai, Yunnan, and Heilongjiang, were in the low-value zone of the renewal capacity toughness level. This indicates that the capacity of China’s logistics industry for renewal maintained a trend of stable development in the last decade but with clear regional differences.

The results of the above spatial analysis based on visualization using ArcGIS show that the degree of resilience of China’s logistics industry in 2012–2021 was high in terms of its capacity for renewal. However, its resistive, adaptive, and restorative capacities were low and more dispersed. They thus need to be improved.

3.1.4. Spatio-Temporal Evolution of Overall Resilience

The geographical arrangement of the comprehensive levels of resilience of the logistics sector in China in 2012, 2015, 2018, and 2021 is shown in Figure 15. It shows that the overall resilience of the logistics sector in all provinces of China increased from 2012 to 2021. But the geographical arrangement of areas with greater degrees of resilience decreased and generally exhibited an arrangement of “low in the northwest and high in the southeast”. In 2012, the comprehensive levels of resilience of the logistics industries in Shandong, Jiangsu, Zhejiang, and Guangdong were in the high-value zone; those of Sichuan, Shaanxi, Hubei, Liaoning, Henan, and Hebei were in the higher-value zone; and the overall levels of resilience of all other provinces were in the low-value zone. Compared with that in 2012, the comprehensive resilience of the logistics industry in Guangdong, Guangxi, Jiangxi, and other southeastern regions did not change significantly in 2015. Shanxi, Anhui, Fujian, Hunan, and other provinces had higher levels of resilience. The comprehensive level of resilience of the logistics sector was in the high-value range in 2021. In terms of the overall distribution, areas with high and medium-to-high levels of comprehensive resilience in the logistics industry were adjacent to one another, while the northern and western regions of China formed a contiguous area that was dominated by low and medium-to-low levels of resilience.

The comprehensive resilience of China’s logistics sector was generally at a medium level. It exhibited a trend of reduction in the quantity of zones with low-level resilience and intensification in areas with high-level resilience. In addition, provinces with a higher level of comprehensive resilience had developed from “monocentric” to “polycentric”. Economically developed provinces had a higher level of comprehensive resilience and served as the center of their agglomeration and contiguous distribution. However, the resilience of provinces in the western and northeastern parts of the country was lower. This shows clear regional differences.

The comprehensive resilience level of China’s three major regions—east, center, and west—was characterized by a decreasing distribution of the “east–center–west” gradient. Across the board, the aggregate level of resilience was slightly higher in the east than in the center and west. The eastern area had an increased degree of economic development, with better infrastructure, public services, and other conditions, such that its resilience to shocks was higher. The development of the logistics industry in western provinces such as Tibet and Xinjiang is susceptible to extreme weather conditions such as drought and snowfall, and is therefore less resilient to shocks. The levels of resilience at different scales showed the characteristics of decreasing economic scales in developed and less developed regions. The combined resilience from east to west and from the coast to inland areas exhibited a gradual decreasing trend.

3.2. Spatial Autocorrelation in the Resilience of China’s Logistics Industry

3.2.1. Results of the Spatial Autocorrelation Test

Based on the global autocorrelation analysis, the toughness value was used as the base data in this paper, and the adjacency matrix was chosen as the spatial weighting file. STATA 17 software was used to calculate the global Moran’s I index of the toughness of China’s logistics industry for the four years of 2012, 2015, 2018, and 2021. In addition, the obtained data were used to perform the spatial autocorrelation test, and the results are shown in

Table 3. Values of the global Moran’s I of the levels of resilience of China’s logistics industry.

Dimension	Year	p-Value	Moran’s I	Z-Score
Resistance	2012	0.053	0.094	1.620
	2015	0.051	0.096	1.632
	2018	0.081	0.078	1.400
	2021	0.011	0.150	2.276
Adaptability	2012	0.037	0.107	1.792
	2015	0.283	0.009	0.575
	2018	0.346	−0.007	0.396
	2021	0.220	0.020	0.771
Resilience	2012	0.002	0.201	2.931
	2015	0.040	0.103	1.747
	2018	0.106	0.058	1.248
	2021	0.113	0.055	1.211
Updating capacity	2012	0.224	0.024	0.758
	2015	0.495	−0.033	0.013
	2018	0.484	−0.032	0.041
	2021	0.445	−0.044	−0.138
Comprehensive resilience	2012	0.031	0.113	1.868
	2015	0.080	0.073	1.406
	2018	0.097	0.057	1.300
	2021	0.052	0.077	1.629

. The values of Moran’s indices for the three dimensions and the composite resilience were all greater than zero, except for the capacity for renewal, indicating a positive spatial autocorrelation. In contrast, the spatial autocorrelation in the capacity for renewal yielded a negative correlation in 2012–2018. The p-values of the resistive and integrative capacities were lower than 0.1, indicating that both capacities passed the test of significance at the 90% level. The p-values of the capacity of the logistics industry to adapt and update were almost all greater than 0.1, indicating that they did not pass the test of significance at the 90% level. From a global perspective, provinces with highly resilient logistics industries were adjacent to one another, as were provinces that had logistics industries with poor resilience. This exhibits a significant clustering effect [33].

The economic variables and the corresponding spatial lag terms are embedded in the axes as horizontal and vertical coordinates. Respectively, we drew Moran’s scatterplot, where the slope of the regression line represents Moran’s I. The Moran scatterplot delineated four sections as follows: the initial quadrant was the area of high-high agglomeration; the second was the region of high-low agglomeration; the third quadrant was the area of low–low agglomeration; and the fourth was the region of low–high agglomeration. We conducted tests of local autocorrelation on data from 2012, 2015, 2018, and 2021 to plot Moran’s scatterplots for each dimension and the combined level of resilience, as shown in Figure 16. The results of the Lisa diagram show that the vast majority of the data fell into the initial and third quadrants. This suggests a significant geographic clustering of the provincial logistics industry’s level of resilience, adaptability, renewal, and overall resilience to risk. They exhibited high–high clustering and low–low clustering, respectively, which is consistent with the findings of the global autocorrelation test.

3.2.2. Kernel Density Estimation of the Resilience of China’s Logistics Industry

(1)

Kernel density estimation of resilience by dimension

The kernel density curve was used in this paper to conduct kernel density analyses of the resilience level of China’s logistics industry in terms of resistance, adaptability, recovery, and renewal. Based on this, we observed the evolutionary trend of resilience in each dimension over the past decade. As is shown in Figure 17, all the dimensions of resilience underwent changes in 2012, 2015, 2018, and 2021. In general, the kernel density curves of resilience along all dimensions shifted slightly to the right each year, but the interval that they spanned did not change much. The magnitude of changes in the gaps among levels was minor. The peak value of the curve of the resilience of the logistics industry decreased while its width increased, indicating that differences in the levels of resilience among provinces increased. The waveform of the curve of the adaptive capacity of the logistics industry also shifted to the right, reflecting an increase in its value and an overall increase in the level of resilience. The peak of the curve of the capacity of the logistics industry for renewal declined and fluctuated. It also exhibited multiple peaks, indicating that differences in the capacity for renewal among provinces increased, and there was a prominent multi-polarization phenomenon.

(2)

Kernel density estimation of the comprehensive resilience of the logistics industry

The trends in the evolution of the kernel density curves of the integrated level of resilience of the logistics industry in 2012, 2015, 2018, and 2021 are shown in Figure 18. The spatial evolution of integrated resilience exhibited a high degree of similarity in these four years but still varied among them. First, the center of the peak of the curve gradually shifted to the right, indicating an overall increase. Second, the peak of the curve had the tendency to decrease in magnitude and increase in width, indicating that differences in the degree of comprehensive resilience among provinces increased. Third, the peak of the curve in 2021 decreased while its width increased. This means that there was a large difference in the level of integrated resilience of the logistics industry among provinces after 2021. The gap among them further increased owing to unsteady and uncoordinated development and the impact of COVID-19 [34].

3.3. Analysis of Factors Influencing the Resilience of the Logistics Industry

The following tests were used in this paper to determine the optimal model to symbolize the resilience of the logistics sector. In the first step, the LM test was used to examine whether there were spatial effects in the model, as displayed in

Table 4. Factors influencing the chosen indices.

Test Method	Results
LM-error	0.860
Robust LM-error	0.077 *
LM-lag	0.002 ***
Robust LM-lag	0.000 ***
Hausman	86.53 ***
LR-SAR	10.3 *
LR-SEM	16.99 ***

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

. The LM-lag and robust LM-lag passed the test of significance at a level of at least 5%, indicating that the residuals estimated by the model were spatially correlated. In the second step, the Hausman test was conducted by using the fixed-effects model, and its p-value led us to reject the first theory with a significance threshold of 1%. This shows that our choice of the fixed-effects model was appropriate. In the third step, we conducted regression by using the spatial lag models (SAR) with time-fixed effects, spatially fixed effects, and spatio-temporally fixed effects. Five factors passed the test of significance at the 1% level, as shown in

Table 5. Results of estimation of the spatial lag model.

Variable		SAR
	Coefficient	t-Value	p-Value
X1	−0.00000000269	−1.01	0.312
X2	−0.00000000117 ***	−4.88	0.000
X3	0.0000000166 ***	9.08	0.000
X4	−0.0000000351 ***	−4.69	0.000
X5	−0.00000000239 ***	−4.98	0.000
X6	0.00000000536 ***	3.98	0.000

Note: *** p < 0.01.

. The capacity for innovation (X3), infrastructure (X6), and the level of resilience of the logistics industry were significantly positively correlated.

4. Discussion

Enhancing the robustness of China’s logistics sector is essential for advancing its sustainable growth. Based on existing research, the logistics industry resilience level measurement model is used in this paper to study the resistance, adaptability, resilience, and renewal of China’s logistics industry. This comprehensive and systematic study of the evolution of spatial and temporal patterns of resilience in the provincial logistics industry from a geographically based perspective enriches research on the body of knowledge about the robustness of the logistics sector’s spatial distribution. The results show that its resilience on the whole was on an upward trend from 2012 to 2021, which is similar to the findings reported by Chen et al. [35]. In the background of rapid economic development, the logistics industry improved its ability to withstand risks and disasters as well as its ability to recover as a whole. However, its resilience in some provinces grew slowly owing to insufficient investment, a lack of infrastructure projects, and other issues. The same situation is also found in other countries. Factors such as limited land resources and difficulties in infrastructure development are important drivers of the slow development of the logistics industry in Japan [36]. The government should seek to accelerate infrastructure projects for logistics and enhance the industrial framework to improve the resilience of the logistics sector [37].

At the provincial level, the resilience level of China’s logistics industry showed significant spatial heterogeneity. Although its overall resilience improved, the corresponding spatial distribution was uneven. It showed a general trend of the east being high, the west being low, and southerly highs and northerly lows. This is in line with Wang et al.’s findings [38]. Guangdong, Jiangsu, and Zhejiang Provinces consistently ranked among the top in terms of the resilience of the logistics industry over the study period, which suggests that they have a sound infrastructure that is resilient to risks. Qinghai, Ningxia Hui Autonomous Region, and Tibet Autonomous Region all had logistics industries with low levels of resilience. These regions suffer from poor economic development and low investment in the logistics industry in terms of capital and technology. This has rendered their logistics industries susceptible to shocks. The relevant government departments should strengthen their guidance for the logistics industry. The central provinces should drive enhancements in the resilience of industries in the surrounding provinces. To improve the overall level of resilience, it is important to focus on the balanced growth of provinces. We need to make sure that the level of resilience of provinces that are economically developed, such as Guangdong, Zhejiang, and Jiangsu, continues to grow steadily. Additionally, the government ought to provide the requisite assistance to the western region and other underdeveloped provinces and give full play to their geographical advantages. Improving their regional resilience would contribute to enhancing their economic development and resistance to crises.

In examining the variables affecting the resilience of the logistics industry, this research found that innovativeness and infrastructure are positively associated with logistics industry resilience at a significance level of 1%. This indicates that they had a considerable influence on the degree of resilience in China’s logistics sector. This is consistent with the results Chen et al. [39]. The ability to innovate helps improve the adaptability and flexibility of the logistics sector so that it can better respond to emergencies and crises. The construction of a complete logistics infrastructure can help improve the logistical carrying capacity, thus enhancing the resilience of the logistics sector. Similarly, factors such as the capacity for logistics innovation and infrastructure development in different regions also have a positive impact on the development of their logistics industry [40]. Thus, the government ought to invest more funds in education in fields related to the logistics industry, improve the system of funding and conditions of the teaching facilities at colleges and universities, and increase the salary and benefits for high-level talent. Through the above measures, we can provide material assurance to enhance the level of resilience of the logistics industry [41].

Inevitably, this study has some limitations. Firstly, the resilience indicators for the logistics industry used in this paper could be improved. There is currently less academic research on resilience in the logistics industry. Comprehensive evaluation indicators for the logistics industry are related to other factors such as natural disasters in addition to resistance, adaptability, resilience, and renewal. Therefore, subsequent studies should strive to find a better indicator system to quantify more factors and make the indicator system more complete. Secondly, the influencing factors of the resilience of the logistics industry need to be further explored. Restricted by the level of research and research time, this paper is not comprehensive enough to analyze the factors influencing the resilience of the logistics industry, and subsequent specific research can be carried out on this issue to better enhance the level of resilience of the logistics industry. Moreover, the analysis can be enriched by providing paths of improvement for the logistics industry. Future research should also seek to solicit the participation of the government and society at all levels to improve the resilience of the logistics industry and to respond to the relevant risks and challenges.

5. Conclusions

In this study, we developed a framework to assess the resilience of the logistics sector in 31 provinces in China based on its resistance, adaptability, and capacity for renewal from 2012 to 2021. The global Moran’s I index and kernel density estimation are used in this paper to characterize the spatio-temporal evolution of China’s logistics industry in terms of its dimensions and the level of composite resilience, as well as to explore its influencing factors. We identified spatial imbalances in the logistics sector’s resilience development in China. This study’s primary conclusions are as follows:

On the one hand, the level of comprehensive resilience of the logistics sector was on an upward trend in China, where the east regions are high, the west regions are low, the south regions are high, and the north regions are low. The overall resilience in coastal regions is high. Examining several aspects of resilience levels, the resistance and adaptive capacity of the logistics industry are low and dispersed and need to be improved. On the other hand, the spatio-temporal distributions of adaptability, resistance, and the capacity for renewal, as well as the overall resilience, exhibited a “gradation” in terms of differentiation. Provinces with more resilient logistics industries formed clusters centered on economically developed regions, while provinces with logistics industries with low resilience were mainly distributed in the central, western, and some eastern regions of China and had prominent regional differences. The capacity for innovation and infrastructure had a significant and positive effect on improving the resilience of China’s logistics sector. Accelerating the construction of the logistics infrastructure, increasing fixed-asset investment in the logistics industry, and upgrading its capacity for innovation and development as well as labor quality are significant approaches to strengthening China’s logistics sector’s resilience.