Assessment of the Regional Vulnerability to Natural Disasters in China Based on DEA Model

Abstract

China is a country highly vulnerable to natural disasters, resulting in significant losses in terms of human casualties, injuries, property damage, economic losses, infrastructure destruction, and so on each year. We propose a conceptual model based on the Data Envelopment Analysis model to evaluate regional vulnerability in mainland China using the annual data of Chinese official statistics from 2006 to 2021. The proposed model includes five input variables: regional total population, per capita GDP, population density, GDP per square kilometer, and regional total fixed investment in water conservancy, environment, and public facilities management. Additionally, it incorporates two output variables: affected people and direct economic loss. The results indicate that the vulnerability level generally decreases from West China through Central China to East China. Based on the new classification method proposed in this study, the regions are divided into five areas. These findings can serve as a reference for policymakers in enhancing disaster planning and improving the efficiency of natural disaster prevention.

1. Introduction

China is one of the countries most heavily affected by natural disasters primarily caused by natural forces or processes, which are beyond human control. These forces can include geological processes (such as earthquakes, volcanic eruptions, or landslides), meteorological phenomena (such as hurricanes, tornadoes, or floods), or climatic events (such as droughts or wildfires). In fact, natural disasters have posed significant threats to life and property safety and have a profound impact on China’s economic and social development, as well as in other countries [1]. For instance, as shown in Figure 1, in 2008, the direct economic loss amounted to CNY 1175.2 billion. In 2013, natural disasters resulted in significant losses, including nearly 390 million people affected and a direct economic loss of CNY 580.5 billion. Furthermore, in 2016, approximately 190 million people were affected by natural disasters, and the direct economic loss exceeded CNY 500 billion.

The Chinese government has been making significant efforts to improve and strengthen disaster prevention and mitigation measures. Disaster preparedness has become a priority on the agenda. The Chinese central government has allocated substantial financial and human resources to support affected individuals and regions. The timely collection of information on natural disasters, such as the number of occurrences, injuries, casualties, and economic losses, is carried out, and this information is promptly released to the public. The impact of natural disasters on regions is a subject of widespread concern and discussion [2,3,4,5,6,7]. However, there is limited discussion on the comparison between regional vulnerability and prevention efficiency.

Figure 1. The main loss caused by natural disasters in mainland China. (Data source: Retrieved from China Statistical Yearbook from 2005 to 2022 [8]).

Figure 1. The main loss caused by natural disasters in mainland China. (Data source: Retrieved from China Statistical Yearbook from 2005 to 2022 [8]).

Through a review of the existing vulnerability-assessment literature, it is observed that many authors have proposed quantitative assessment methods [9,10,11,12,13,14,15,16]. Nevertheless, the majority of these approaches quantify regional vulnerability through the creation of composite indices and the calculation of sub-indices, using techniques such as artificial neural networks and analytic hierarchy processes [17,18,19,20]. Consequently, the results of vulnerability assessments heavily depend on indicator selection and the weight assigned to the sub-indices, which may undermine their credibility.

Our study aims to evaluate regional vulnerability to natural disasters using a Data Envelopment Analysis (DEA)-based model that avoids the creation of complex indices and the calculation of sub-indices. The framework shown is Figure 2 includes six main steps: literature review, data collection and processing, input and output variables specification, DEA-BCC model setup, efficiency calculation, and efficiency analysis. On the basis of the literature, we construct an input–output model, select mainland China as a case study, and employ data from annual governmental statistics spanning from 2006 to 2021 to assess regional vulnerabilities. Finally, we evaluate and analyze regional vulnerabilities from both temporal and spatial perspectives. It is hoped that this study will provide valuable insights for comprehensive regional disaster management planning, thereby enhancing regional resilience to natural disasters.

2. Literature Review

The concept of vulnerability was first introduced by O’Keefe et al. (1976) [21]. The terminology associated with vulnerability varies among disciplines and research areas [22,23,24,25,26]. The concept of vulnerability is frequently utilized in the context of risk, hazards, and disaster literature [1,27,28]. Since the 1980s, the terms “vulnerable” and “vulnerability” have been increasingly used in disaster-related literature [29]. The United Nations Office for Disaster Risk Reduction (UNDRR) defines vulnerability as the set of conditions determined by physical, social, economic, and environmental factors or processes that increase the susceptibility to the impacts of hazards [30]. In general, natural disasters are perceived as the intersection of natural hazards and vulnerable conditions, and the risk of disaster is high when one or more natural hazards occur in a vulnerable situation [31]. Vulnerability serves as a measure of a region’s capacity to withstand disasters, taking into account the unique economic characteristics of different regions [27]. Cutter and Finch (2008) [32] examined social vulnerability to natural disasters by considering temporal and spatial changes. In this study, vulnerability is employed to assess the capacity to withstand and recover from disasters in regions characterized by distinct economies.

Conducting vulnerability analyses of different regions is crucial and valuable, as they can offer guidance to governments in allocating relief funds and assist regional authorities in enhancing their disaster response capabilities [33]. In recent years, vulnerability assessment has received increasing attention in scientific literature, leading to the development of various conceptual models and assessment methods for evaluating vulnerability [33,34,35]. Among these literatures, the analysis of regional vulnerability has been particularly emphasized. However, given the intricate nature of natural disasters, vulnerability analysis and regional vulnerability assessment often involve subjectivity, such as expert judgment and the weighting of various factors, as the outcomes of these assessments are dependent on the chosen methodology [33].

Traditional methods for regional vulnerability assessment primarily involve the establishment of indicator systems, wherein key factors are selected to reflect the vulnerability conditions of a region. The weights of each evaluation indicator are determined using techniques such as artificial neural networks, analytic hierarchy process (AHP), or principal component analysis (PCA), followed by the calculation of a vulnerability index [36,37]. For instance, Ghahroudi et al. (2016) [36] employed AHP and fuzzy functions to analyze flood vulnerability and created a vulnerability map of the northwestern areas of Tehran, Iran. Kazakis et al. (2015) [18] introduced a multi-criteria index and utilized AHP to calculate parameter weights, thus assessing flood hazard areas at a regional scale. Romero-Lankao et al. (2016) [38] combined a fuzzy logic approach with AHP to examine the relative importance of the selection of indicators in assessing vulnerability to climate hazards in Mumbai, India. However, the weight values obtained are sensitive to indicators for assessment and the weighting methods utilized.

Data Envelopment Analysis (DEA) is a relatively recent “data-oriented” approach used to assess the performance of a group of comparable entities known as Decision Making Units (DMUs), which convert multiple inputs into multiple outputs [39]. In comparison with other commonly used methods, DEA is a fractional programming model that can accommodate multiple inputs and outputs without requiring the explicit specification of functional relationships between them, eliminating the need for predefined weights [40]. Since its introduction in 1978 by Charnes et al. [39], the DEA model has gained significant recognition as a robust methodology and has been extensively employed to evaluate the efficiency of organizational units in various domains, including local government departments, hospitals, schools, banks, retail stores, manufacturing, and similar settings where a relatively homogeneous set of units exists [16].

Considering the complexity of the disaster system and the efficiency of DEA, we aim to utilize the DEA model to investigate regional vulnerability to natural disasters. Previous studies applying this approach in the field of disaster management are limited [9,33,41]. Huang et al. (2011) [9] developed an input–output DEA model to assess the regional vulnerability of mainland China. They established a comprehensive evaluation indicator system that considered regional hazard risks, exposure of the regional socioeconomic system, and regional disaster losses. Principal component analysis was used to extract key factors for assessing regional vulnerability. Zhou et al. (2014) [41] also employed this technique to evaluate the social vulnerability of Chinese mainland provinces to natural hazards. They divided vulnerability into socioeconomic and built-environmental vulnerability, identified principal factors using factor analysis, and assessed provincial vulnerability values for 2000 and 2010. Wei et al. (2004) [33] argued that traditional methods were unreliable due to their sensitivity to weight selection for the sub-indices used to calculate composite indices for regional vulnerability. They employed the DEA model with two inputs (population density and complexity of commercial infrastructure) and two outputs (number of people affected by the disaster and total cost of damage) to evaluate regional vulnerability based on official annual data. They proposed a novel method to classify regional vulnerability to natural disasters. While there are numerous measurable indices, based on a literature review, the frequently discussed factors for regional vulnerability include the number of disasters, regional population, GDP, per capita GDP, population density, GDP per square kilometer, fixed investment in construction, direct economic loss, and affected population. However, previous studies have some limitations, such as subjective indicator selection and determination, or the omission of key factors. In this article, we construct an input–output DEA-based model to evaluate the regional vulnerabilities of mainland China and compare them temporally and spatially.

3. Data and Methodology

3.1. Data

The investigated region in our study focuses on mainland China. Specifically, we examined the regional vulnerabilities of its 31 provincial regions, which consist of 22 provinces, 5 autonomous regions, and 4 municipalities.

In order to evaluate regional vulnerabilities, we carefully considered the existing literature review and data availability. As a result, we selected seven factors as input–output variables for our analysis. These factors include the number of affected people, direct economic loss, regional total population, per capita GDP, population density, GDP per square kilometer, and regional total fixed investment in water conservancy, environment, and public facilities management.

To ensure the accuracy and reliability of our findings, the data used in our study were derived from annual official statistics of China, covering the period from 2006 to 2021. These statistics provide a comprehensive and representative source of information for assessing the regional vulnerabilities of mainland China.

3.2. Methodology

DEA model is an effective method for evaluating the relative efficiencies of different decision-making units (DMUs) that are homogeneous in nature. It is also flexible enough to handle systems with multiple inputs and outputs. Therefore, we employed the DEA model to assess regional vulnerabilities. Based on a review of the literature, the model incorporated five inputs and two outputs. The inputs consisted of regional total population, per capita GDP, population density, GDP per square kilometer, and regional total fixed investment in water conservancy, environment, and public facilities management. The outputs were affected people and direct economic loss. The relative efficiency derived from the DEA model reflected the sensitivity of a region to natural disasters. In the context of vulnerability to natural disasters, higher index values indicate greater vulnerability, meaning that the region is more susceptible to natural disasters or experiences more damage at the same level of natural disasters. Conversely, lower index values indicate lower vulnerability.

Following the CCR model proposed by Charnes, Cooper, and Rhodes [39], various DEA models have been introduced in the literature, including the BCC model [42] and the multiplicative model [43]. Among these, the CCR and BCC models are the most widely discussed and commonly used. The CCR model assumes constant returns to scale (CRS), while the BCC model extends the CCR model by considering variable returns to scale (VRS). The BCC model distinguishes between technical inefficiency and scale inefficiency, allowing for an assessment of pure technical efficiency and identifying the presence of increasing, decreasing, or constant returns to scale possibilities [42].

In this study, we employed the DEA-BCC model for data analysis using the DEAP Version 2.1 software [44]. We assume that regional vulnerability is positively correlated with the losses caused by natural disasters. The DEA-BCC model is used to evaluate the vulnerability of each region in different years and to determine the comprehensive vulnerabilities of each region from 2006 to 2021. We consider a set of n decision-making units, denoted as DMU_j (where j = 1, 2,…, n). Each decision-making unit is characterized by m inputs and s outputs. The input-oriented BCC model, developed by Banker, Charnes, and Cooper [42], can be described as follows:

$$\left\{\begin{array}{c}min\left[\theta -\epsilon e^T\left(s^{-}+s^{+}\right)\right]\hfill \\ {\displaystyle \sum _{j=1}^n}{\lambda }_j\left(j_0\right)x_{ij}+s^{-}=\theta x_{i0},i=1,2,\cdots ,m\hfill \\ {\displaystyle \sum _{j=1}^n}{\lambda }_j{y}_{rj}-s^{+}=y_{r0},r=1,2,\cdots ,s\hfill \\ {\displaystyle \sum _{j=1}^n}{\lambda }_j=1\hfill \\ \epsilon ,{\lambda }_j\left(j_0\right),s^{-},s^{+}\ge 0,j=1,2,\cdots ,n\hfill \end{array}\right.$$

(1)

In the DEA-BCC model, x_ij (i = 1,…, m) represents the i-th input of the j-th DMU_j (j = 1, 2,…, n), and y_rj (r = 1,…, s) represents the r-th output. In this study, we considered five inputs: x_1i, x_2i, x_3i, x_4i, and x_5i, which correspond to the regional total population, per capita GDP, population density, GDP per square kilometer, and regional total fixed investment in water conservancy, environment, and public facilities management for the i-th region, respectively. The two outputs, y_1i and y_2i, represent the number of affected people and the direct economic loss caused by natural disasters in the i-th region (i = 1, 2,…, n) or DMU.

The efficiency score, θ (0 < θ ≤ 1), in DEA terms reflects the regional vulnerability to natural disasters. A smaller θ indicates a lesser impact on the i-th region, whereas a larger θ indicates a higher vulnerability. λ_j (j = 1, 2,…, n) represents the weight value assigned to the j-th region. ε is a non-Archimedean infinitesimal value smaller than any positive real number. The S⁻ and S⁺ variables are slack variables, and ${\displaystyle \sum }_{j=1}^n{\lambda }_j$ represents the variable returns to scale constraints.

In this study, we utilized the original data collected by the Chinese government to calculate the annual vulnerability efficiency scores. Since the losses caused by natural disasters exhibit significant variations across regions over the years, and the vulnerabilities to natural disasters also show random variations in different years, we aim to gain a comprehensive understanding of regional vulnerabilities from 2006 to 2021. Therefore, we apply the DEA-BCC model to compute the overall regional vulnerabilities using the annual data spanning from 2006 to 2021.

4. Empirical Analyses

The main objective of this study was to evaluate the regional vulnerabilities using the results obtained from the DEA-BCC model. To achieve this, we employed five input variables and two output variables to calculate the efficiencies of the 31 decision-making units (DMUs). Through data processing and analysis, we obtained annual and comprehensive regional efficiencies for each region, as presented in

Table 1. Index of annual vulnerabilities to natural disasters for each region in mainland China from 2006 to 2021.

DUM	Province (Municipality)	2006	2007	2008	2009	2010	2011	2012	2013	2014	2015	2016	2017	2018	2019	2020	2021
1	Beijing	0.218	0.211	0.220	0.228	0.233	0.247	0.267	0.275	0.290	0.300	0.295	0.590	0.394	0.334	0.440	0.553
2	Tianjin	0.303	0.288	0.248	0.246	0.239	0.241	0.523	0.265	0.278	0.296	0.301	0.411	0.383	0.503	0.489	0.478
3	Hebei	0.594	0.706	0.462	0.848	0.480	0.493	1.000	0.619	0.768	1.000	1.000	0.896	0.655	0.734	0.744	0.768
4	Shanxi	0.728	1.000	0.571	0.646	0.570	0.592	0.639	0.695	0.743	0.879	0.812	0.975	0.732	1.000	0.984	1.000
5	Inner Mongolia	1.000	1.000	0.779	1.000	0.956	1.000	1.000	0.803	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
6	Liaoning	0.460	1.000	0.386	0.714	0.366	0.361	0.833	0.399	0.461	0.592	0.668	1.000	1.000	1.000	1.000	1.000
7	Jilin	0.548	0.833	0.526	0.953	1.000	0.496	0.594	0.536	0.584	0.627	0.609	1.000	0.647	0.837	0.922	0.812
8	Heilongjiang	0.629	1.000	0.571	0.723	0.595	0.603	0.618	0.662	0.679	0.682	1.000	1.000	0.817	1.000	0.941	0.884
9	Shanghai	0.187	0.195	0.209	0.217	0.223	0.240	0.263	0.288	0.295	0.307	0.408	0.590	0.416	0.330	0.325	0.356
10	Jiangsu	0.352	0.378	0.291	0.291	0.276	0.283	0.412	0.325	0.318	0.315	0.312	0.433	0.287	0.285	0.298	0.300
11	Zhejiang	0.509	1.000	0.631	0.438	0.300	0.393	1.000	0.703	0.368	1.000	0.401	0.218	0.351	1.000	0.370	0.393
12	Anhui	0.602	1.000	0.741	1.000	0.619	0.664	0.852	0.876	0.765	0.782	1.000	0.365	0.656	0.712	1.000	0.585
13	Fujian	1.000	0.503	0.417	0.416	0.786	0.394	0.417	0.430	0.433	0.996	0.870	0.387	0.404	0.468	0.401	0.403
14	Jiangxi	0.862	0.964	0.833	0.743	0.653	0.654	0.736	0.763	0.743	0.711	0.709	0.636	0.660	1.000	1.000	0.680
15	Shandong	0.371	0.572	0.323	1.000	0.327	0.417	0.585	0.428	0.426	0.582	0.412	1.000	0.722	1.000	0.502	0.503
16	Henan	0.462	0.687	0.526	0.864	0.520	0.552	0.656	0.773	1.000	0.667	0.685	0.929	1.000	0.855	0.799	1.000
17	Hubei	0.601	0.791	0.910	0.658	0.497	0.537	0.646	0.688	0.553	0.694	1.000	0.838	0.743	0.709	1.000	0.558
18	Hunan	1.000	1.000	1.000	1.000	0.565	1.000	0.659	0.867	0.822	0.669	0.904	1.000	0.591	0.825	0.742	0.624
19	Guangdong	1.000	0.321	0.295	0.307	0.307	0.328	0.395	0.626	0.436	1.000	0.410	1.000	0.382	0.354	0.412	0.419
20	Guangxi	1.000	0.791	0.886	0.822	0.686	0.691	0.747	0.777	0.794	0.758	0.737	0.354	0.754	0.819	0.819	0.837
21	Hainan	0.796	1.000	1.000	0.769	0.709	1.000	0.713	0.771	1.000	0.883	1.000	0.864	0.794	0.811	0.879	0.843
22	Chongqing	1.000	1.000	0.612	0.593	1.000	0.501	0.611	0.589	0.584	0.548	0.535	0.857	0.550	0.529	0.676	0.538
23	Sichuan	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.774	0.700	0.680	1.000	1.000	1.000	0.824
24	Guizhou	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.880	0.907	0.510	0.760	0.739	0.781	0.809
25	Yunnan	0.951	1.000	0.962	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	0.844	1.000	1.000	1.000
26	Tibet	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
27	Shananxi	0.832	1.000	0.593	0.596	0.552	0.609	0.562	0.624	0.751	0.599	0.574	1.000	0.539	0.629	0.572	0.950
28	Gansu	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000	1.000
29	Qinghai	1.000	1.000	0.772	0.957	0.916	1.000	1.000	0.771	1.000	1.000	0.943	1.000	0.937	1.000	1.000	1.000
30	Ningxia	0.903	1.000	0.774	0.794	0.633	0.662	0.771	0.665	0.821	0.790	0.813	1.000	0.785	0.866	0.999	1.000
31	Xinjiang	0.758	0.905	1.000	0.790	1.000	0.784	1.000	0.700	1.000	1.000	1.000	1.000	0.955	1.000	0.998	0.979

.

4.1. Regional Analysis of the Annual Vulnerabilities to Natural Disasters

Table 1 provides an overview of the annual regional vulnerabilities to natural disasters from 2006 to 2021. Here, we take the year 2006 as an example to analyze the results.

The regions most severely affected by natural disasters in 2006 were Inner Mongolia Autonomous Region, Fujian Province, Hunan Province, Guangdong Province, Guangxi Province, Chongqing Province, Sichuan Province, Guizhou Province, Tibet Autonomous Region, Gansu Province, and Qinghai Province, with vulnerability indices equal to 1.

The regions that were significantly impacted by natural disasters in 2006 were Jiangxi Province (efficiency score of 0.826), Hainan Province (0.796), Yunnan Province (0.951), Shananxi Province (0.832), Ningxia Hui Autonomous Region (0.903), and Xinjiang Uygur Autonomous Region (0.758).

Shanghai municipality was the region least negatively impacted by natural disasters in 2006, with an efficiency score of 0.187. Beijing Municipality, Tianjin Municipality, Jiangsu Province, and Shandong Province were lightly affected by natural disasters in 2006, with efficiency indices of 0.218, 0.303, 0.352, and 0.371, respectively.

It is worth noting that some regions consistently maintain lower vulnerability indices over the years, such as Beijing, Tianjin, and Shanghai. This suggests that these regions have higher capacities to withstand natural disasters.

Additionally, there are regions whose efficiency scores show significant variation across different years, such as Hebei Province, Liaoning Province, Zhejiang Province, Anhui Province, Fujian Province, Shandong Province, Guangdong Province, and Tibet Autonomous Region (with θ ranging from 0.2 to 1). This indicates a fluctuation in their vulnerability levels over time. Moreover, the efficiency scores also vary considerably across different geographic regions, highlighting the spatial variation in vulnerability to natural disasters.

Overall, the results in Table 1 demonstrate the regional vulnerabilities to natural disasters in different years and provide valuable insights for understanding the dynamics and spatial patterns of vulnerability across China’s mainland.

Table 2. The most severely affected regions by natural disasters from 2006 to 2021.

Year	The DUM
2006	Inner Mongolia, Fujian, Hunan, Guangdong, Guangxi, Chongqing, Sichuan, Guizhou, Tibet, Gansu, Qinghai
2007	Shanxi, Inner Mongolia, Liaoning, Heilongjiang, Zhejiang, Anhui, Hunan, Hainan, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shananxi, Gansu, Qinghai, Ningxia
2008	Hunan, Hainan, Sichuan, Guizhou, Tibet, Gansu, Xinjiang
2009	Inner Mongolia, Anhui, Shandong, Hunan, Sichuan, Guizhou, Yunnan, Tibet, Gansu
2010	Jilin, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Gansu, Xinjiang
2011	Inner Mongolia, Hunan, Hainan, Sichuan, Guizhou, Yunnan, Tibet, Gansu, Qinghai
2012	Hebei, Inner Mongolia, Zhejiang, Sichuan, Guizhou, Yunnan, Tibet, Gansu, Qinghai, Xinjiang
2013	Sichuan, Guizhou, Yunnan, Tibet, Gansu
2014	Inner Mongolia, Henan, Hainan, Sichuan, Guizhou, Yunnan, Tibet, Gansu, Qinghai, Xinjiang
2015	Hebei, Inner Mongolia, Zhejiang, Guangdong, Yunnan, Tibet, Gansu, Qinghai, Xinjiang
2016	Hebei, Inner Mongolia, Heilongjiang, Anhui, Hubei, Hainan, Yunnan, Tibet, Gansu, Xinjiang
2017	Inner Mongolia, Liaoning, Jilin, Helongjiang, Shandong, Hunan, Guangdong, Yunnan, Tibet, Shananxi, Gansu, Qinghai, Ningxia, Xinjiang
2018	Inner Mongolia, Liaoning, Henan, Sichuan, Tibet, Gansu
2019	Shanxi, Inner Mongolia, Liaoning, Heilongjiang, Zhejiang, Jiangxi, Shandong, Sichuan, Yunnan, Tibet, Gansu, Qinghai, Xinjiang
2020	Inner Mongolia, Liaoning, Anhui, Jiangxi, Hubei, Sichuan, Yunnan, Tibet, Gansu, Qinghai
2021	Shanxi, Inner Mongolia, Liaoning, Henan, Yunnan, Tibet, Gansu, Qinghai, Ningxia

provides a list of regions with annual efficiency scores of 1 (θ = 1) from 2006 to 2021. By examining the results in Table 2, the following observations can be made:

There are 27 regions that have experienced at least 1 year with an efficiency score of 1 during the 16-year period. This indicates that natural disasters occur randomly and that there is a need for these regions to improve their capacities to withstand such disasters.

The Tibet Autonomous Region and Gansu Province rank first in terms of the occurrence frequency, with an efficiency score of 1 appearing sixteen times each. They are followed by Yunnan Province and Inner Mongolia Autonomous Region, which have an efficiency score of 1 appearing thirteen times each. Qinghai Province appears ten times with an efficiency score of 1. Sichuan Province and Guizhou Province both have an efficiency score of 1 appearing nine times each, while Xinjiang Uygur Autonomous Region appears eight times.

The most affected regions primarily lie in the western region of mainland China. Fewer regions in the central and eastern regions of China appear in Table 2, suggesting that the western regions are more prone to natural disasters or have higher vulnerability levels.

These findings highlight the regional disparities in terms of vulnerability to natural disasters within China’s mainland, with the western regions being more frequently impacted. It underscores the importance of enhancing disaster resilience and preparedness in these regions to minimize the impacts of future natural disasters.

4.2. Regional Analysis of the Comprehensive Vulnerabilities

Table 3. Overall regional vulnerability efficiency to natural disasters in mainland China (2006–2021).

DUM	Province (Municipality)	District	Efficiency Score	Vulnerability Rank	DUM	Province (Municipality)	District	Efficiency Score	Vulnerability Rank
1	Beijing	East	0.218	Slight	17	Hubei	Central	0.601	Medium
2	Tianjin	East	0.303	Low	18	Hunan	Central	1.000	Extreme
3	Hebei	East	0.594	Medium	19	Guangdong	East	1.000	Extreme
4	Shanxi	Central	0.728	Medium	20	Guangxi	West	1.000	Extreme
5	Inner Mongolia	West	1.000	Extreme	21	Hainan	East	0.796	Severe
6	Liaoning	East	0.460	Low	22	Chongqing	West	1.000	Extreme
7	Jilin	Central	0.548	Medium	23	Sichuan	West	1.000	Extreme
8	Heilongjiang	Central	0.629	Medium	24	Guizhou	West	1.000	Extreme
9	Shanghai	East	0.187	Slight	25	Yunnan	West	0.951	Extreme
10	Jiangsu	East	0.352	Low	26	Tibet	West	1.000	Extreme
11	Zhejiang	East	0.509	Medium	27	Shananxi	West	0.832	Severe
12	Anhui	Central	0.602	Medium	28	Gansu	West	1.000	Extreme
13	Fujian	East	1.000	Extreme	29	Qinghai	West	1.000	Extreme
14	Jiangxi	Central	0.862	Severe	30	Ningxia	West	0.903	Extreme
15	Shandong	East	0.371	Low	31	Xinjiang	West	0.758	Severe
16	Henan	Central	0.462	Low

and Figure 3 present the results of the comprehensive regional vulnerability analysis based on the DEA-BCC model. The analysis utilizes panel data consisting of 434 data points, representing annual data for the 31 regions over a 16-year period.

The regional vulnerabilities are classified into five degrees based on the relative efficiency scores (θ) obtained from the DEA-BCC model. These degrees are categorized as slight (0 < θ ≤ 0.25), low (0.25 < θ ≤ 0.5), medium (0.5 < θ ≤ 0.75), severe (0.75 < θ ≤ 0.9), and extreme (0.9 < θ ≤ 1).

Table 3 provides a comprehensive overview of the regional vulnerabilities for each region and year, indicating the degree of vulnerability based on the corresponding θ values. Figure 3 visualizes the regional vulnerabilities across the 31 DUMs, allowing for a clearer understanding of the distribution and variations in vulnerability levels over the 16-year period.

These results provide valuable insights into the overall vulnerability landscape in China’s mainland, highlighting regions that consistently exhibit high vulnerability levels and those that experience fluctuations in vulnerability over time. This information can assist policymakers and stakeholders in prioritizing resources, implementing mitigation measures, and developing strategies to enhance resilience in regions prone to natural disasters.

Figure 3. Regional vulnerability index based on data from 2006 to 2021.

Figure 3. Regional vulnerability index based on data from 2006 to 2021.

Based on the information provided in Table 3, we can observe the following patterns and trends in regional vulnerabilities to natural disasters in China’s mainland:

The regions with slight vulnerability are Beijing, Tianjin, Shanghai, and Hainan Province, located in the eastern region of China, which consistently exhibit lower vulnerability levels compared with other regions. This suggests that these areas have relatively higher capacities to withstand natural disasters.

The regions with low vulnerability are Tianjin, Liaoning Province, Jiangsu Province, Shandong Province, and Henan Province, which are primarily located in East China or Central China and demonstrate relatively low vulnerability levels. These regions have a better overall performance in terms of vulnerability.

The regions with severe vulnerability are Jiangxi Province, Hainan Province, Shaanxi Province, and Xinjiang Uygur Autonomous Region, representing the eastern, central, and western regions of China, respectively, which exhibit higher vulnerability levels. These regions require greater attention and targeted measures to mitigate the impacts of natural disasters.

Table 3 indicates that the majority of regions classified under the extreme vulnerability level are located in West China, accounting for 76.9% of the total. This highlights the higher vulnerability levels in the western regions and the need for comprehensive strategies to address and reduce their vulnerability.

These findings emphasize the regional disparities in vulnerability across China’s mainland. The eastern regions generally display lower vulnerability, while the western regions face greater challenges. This information can aid in the formulation of policies and measures to enhance resilience and reduce vulnerabilities in the regions most prone to natural disasters.

The regional analysis of annual and comprehensive vulnerabilities to natural disasters reveals distinct patterns across different regions in China. The regions with slight and low vulnerability are predominantly located in the eastern region, while regions with severe and extreme vulnerability are concentrated in the central and western regions, with some exceptions like Fujian province, Guangdong province, and Hainan province. West China consistently shows high vulnerability levels with fluctuations, while East China generally exhibits lower and more stable vulnerability levels. The vulnerability in Central China experiences significant variations.

This information can serve as valuable references for government decision-making in natural disaster management. It highlights the need to prioritize resources and allocate investments in regions with higher vulnerability, particularly in the central and western regions. These regions face challenges due to unfavorable natural factors and comparatively lower levels of total fixed investment in water conservancy, environment, and public facilities management. The disparities in relief funds distribution also contribute to the overall higher vulnerability from East China to West China.

The approach used in this study differs from traditional methods that rely on complex indicator systems and weights determined through techniques such as artificial neural networks. Instead, the study utilizes original data from official statistical organizations to evaluate relative vulnerability. The proposed classification criteria for regional vulnerability in China provide a new framework for understanding and categorizing vulnerabilities. By considering the specific characteristics of each region, policymakers can devise targeted strategies and allocate resources effectively to reduce vulnerabilities and enhance resilience to natural disasters.

5. Discussion

In this study, we have presented a vulnerability analysis model to evaluate regional vulnerabilities in mainland China. The model utilizes an input–output DEA approach and relative vulnerability indices based on Chinese official statistical data from 2006 to 2021. Unlike previous quantitative methods that employ complex indicator systems and various techniques to handle indicators [7,36,37], our approach is simpler and relies on original Chinese official statistical data, which enhances the credibility of the results.

We have classified regional vulnerabilities into different levels, providing a new classification method for regional vulnerability assessment. This classification allows for a better understanding of the variations in vulnerability across different regions in mainland China.

The findings of our study reveal distinct patterns in regional vulnerabilities. The eastern region generally exhibits lower vulnerability levels, while the central and western regions face higher vulnerability, with some exceptions. This information is valuable for decision-making in disaster management planning, as it highlights the need for targeted strategies and resource allocation to address the specific vulnerabilities of each region.

The analysis conducted in this study using the DEA model reveals significant regional disparities in vulnerability within mainland China. There is a clear negative correlation between regional economic development and vulnerability levels, with the western region exhibiting the highest vulnerability, followed by the central region and the eastern region. The analysis over the 16-year period from 2006 to 2021 also highlights the relatively stable vulnerability in East China and West China, while the regions in Central China experience more significant fluctuations in vulnerability levels over time.

It is worth noting that vulnerability is not only inversely related to regional economic development but also negatively correlated with the level of total fixed investment and disaster preparedness. More developed regions tend to have lower vulnerability levels due to their greater capacity to withstand natural disasters. However, it is important to acknowledge the exceptions of Fujian Province, Guangdong Province, and Hainan Province. Despite being located on the south coast of China and experiencing rapid economic development, these provinces are densely populated and are frequently affected by land-falling tropical cyclones, which contributes to their higher vulnerability levels.

The findings of this study emphasize the importance of considering both economic development and specific geographical factors when assessing regional vulnerability. Policymakers and stakeholders can use these insights to prioritize disaster management efforts and allocate resources effectively, taking into account the unique vulnerabilities of different regions. By addressing these vulnerabilities, especially in high-risk areas, the resilience and preparedness of these regions can be strengthened to mitigate the impacts of natural disasters.

Future studies can further explore how regions can enhance preparedness and improve resilience in the face of natural disasters. Examining preparedness measures in regions with low vulnerability can reveal effective strategies for reducing disaster risks. Identifying factors contributing to vulnerability in high-risk regions can guide targeted interventions. Additionally, we have future plans to explore the vulnerability and resilience specific to certain types of natural disasters, with a particular focus on earthquakes. Building upon our previous work on ambient seismic noise for urban disaster management [45], this investigation aims to uncover unique vulnerabilities associated with seismic events and contribute to the development of tailored strategies.

Overall, these research efforts contribute to a better understanding of vulnerability, enhancing disaster preparedness, and promoting resilience.

6. Conclusions

In conclusion, this vulnerability analysis model, based on input–output DEA and relative vulnerability indices, provides a robust and convincing approach to assessing regional vulnerabilities in mainland China. The proposed classification method offers a new perspective on regional vulnerability assessment. The results of our study can assist policymakers in developing effective disaster management plans and allocating resources to reducing vulnerabilities and enhancing resilience in different regions.

The findings of this study contribute to a comprehensive understanding of regional vulnerability in mainland China. The input–output DEA model, coupled with the use of original official statistics, enables the generation of detailed and reliable regional vulnerability assessments. The proposed classification method provides policymakers with valuable references for promoting disaster planning and implementing effective prevention measures.

It is important to acknowledge that while natural disasters are beyond human control, human actions can play a significant role in disaster prevention and mitigation efforts. By exploring areas such as preparedness measures, factors contributing to vulnerability, and the effectiveness of disaster management practices (e.g., [46,47]) in future studies, policymakers and stakeholders can gain a deeper understanding of how to enhance regional preparedness, improve disaster response, and build resilient communities to minimize the adverse impacts of natural disasters.