Livelihood Resilience of Rural Residents under Natural Disasters in China

Abstract

The impact of natural disasters on rural areas has recently increased, the question of how to effectively improve the livelihood resilience of rural residents has therefore become an essential issue. In this context, a livelihood resilience evaluation index system for rural residents was constructed in four dimensions: livelihood quality, livelihood promotion, livelihood provision, and disaster stress. A structural dynamics model was used to analyze the changing characteristics of the livelihood resilience of rural residents in China between the years 1980 and 2020, and a livelihood resilience trend from 2021 to 2030 was predicted based on the ARIMA model. The main factors influencing livelihood resilience were explored using ridge regression analysis. The results show that: (1) livelihood resilience of rural residents in China fluctuated significantly between 1980 and 2020, tending generally to increase; (2) livelihood resilience is positively correlated with livelihood quality, livelihood promotion, and livelihood provision, while it is negatively correlated with disaster stress; (3) livelihood quality, livelihood promotion, and livelihood supply will still increase between the years 2021 and 2030, while the increase of livelihood resilience will tend to slow down; and (4) six variables have a significant positive impact on livelihood resilience, and provide a basis for the subsequent enhancement of livelihood resilience.

1. Introduction

Since the 21st century, the world has faced serious impacts and challenges caused by climate change and extreme weather. Historical observational data show that climate change has led to an increase in the frequency and intensity of extreme weather events (heavy rainfall, drought, hail, floods, frost, etc.) in many regions [1]. The rural area is a complex system with dynamic, open, and regional characteristics [2]. With the transformation and upgrading of society, the flow of factors between urban and rural areas has become increasingly complex, and uncertain factors in rural areas have greatly increased. Phenomena and problems such as population aging, labor outflow, land abandonment and hollow villages are gradually emerging, and the rural recession caused by these problems have become a global trend [3,4,5]. Under the influence of the COVID-19 pandemic, rural areas have been more impacted, making the research of rural livelihood resilience very urgent. Meanwhile, rural residents, as the most basic livelihood units and actors in the rural system, are the most direct bearers of climate disturbances. Their livelihood resilience is directly or indirectly affected by external threats and shocks such as natural disasters [6]. As a result, the sensitivity to climate change of agriculture and rural residents who depend on natural resources is changing. However, due to the different ability of rural residents’ livelihood systems to resist the disturbance of potential disasters, the livelihood elements and functional structure of rural areas will also change accordingly, making those livelihoods vulnerable. Therefore, the question of how to prevent or mitigate the impact of natural disasters caused by climate change on livelihood resilience, especially rural households’ livelihood resilience, has gradually become an important issue that needs to be solved urgently around the world.

“Resilience” comes from the Latin word “resilio”, which means to return to the original state. With the deepening of studies on resilience, the concept has been discussed by many scholars from different disciplines, such as physics, engineering, ecosystems, disaster management, and organizational research [7,8,9,10]. According to the relevant definition of resilience by scholars such as Holling, Walker and Folke, the current “resilience” theory is considered to be a theory of a system’s response to extreme disturbances and continuous stress arrived at by analyzing the ability of a system to absorb, cope with, and adapt to changes after being disturbed [11]. It can provide new tools and concepts for ecological environment, natural resource management and socio-economic sustainable development research [12]. Many scholars, especially in the face of the uncertainties brought about by future climate change, believe that resilience thought could be the optimal way to enhance livelihoods and promote sustainable development [13,14,15]. “Livelihood”, as a way of human behavior, is a means of survival, living and production, which is crucial to the sustainable development of society and its individuals. As for the definition of livelihood, Scoones believes that livelihood consists of the capabilities, assets and actions needed for life [16], while Ellis believes that livelihood is a way of survival, which is closely related to income, because income reflects the availability of survival needs to a certain extent, rather than just the result of them [17].

Livelihood resilience is the combination of the concepts and connotations of resilience and livelihoods, which are summarized and applied to all levels of society to explore the resilience of subjects at different levels after disturbances and shocks [18]. It has been widely applied to various directions of research on climate change, disaster management, food security, social ecology and sustainable development [19,20,21,22]. In terms of research perspectives and scales, scholars have conducted studies on livelihood resilience at macro, meso, and micro levels, covering different scales of research such as national, regional, community, and household [23,24]. In terms of research objects, scholars have also shifted their focus from livelihood resilience and livelihood survival to livelihood recovery, learning and adaptive capacity in the context of policy implementation or climate change [25]. In terms of evaluation methods, most of the methods currently used involve sociology, statistics, mathematics and other disciplines, mainly using qualitative and quantitative analyses. Qualitative analysis includes questionnaires, participant observation, and semi-structured interviews [26,27,28], but its livelihood resilience evaluation is more subjective. Quantitative analysis includes entropy weighting method, principal component analysis, and functional modeling method [29,30,31], but the nonlinear relationship between elements is ignored to some extent [32].

It is worth noting that although some research results have been achieved, the following issues still need to be discussed. Firstly, existing studies lack long time series data, especially tracking studies on long time data of more than forty years. There is insufficient attention to the dynamics of livelihoods, and the understanding of livelihood evolution process and evolution mechanism is still limited [33]. Secondly, although climate change is closely related to the current livelihood resilience of rural residents, few studies have been conducted to integrate them, and there is a lack of research on the coupling relationship and impact mechanisms between them. Finally, existing rural residents’ livelihood resilience assessments are often based on comprehensive household surveys, in-depth structured interviews, and comparative analysis of data on changes in rural livelihood capital or livelihood strategies [34,35]. It is difficult to reflect the dynamic characteristics of livelihood resilience and the internal structural effects affecting livelihood resilience, and there is a lack of dynamic analysis and empirical research on changes in farmers’ livelihood resilience. In view of the above, coupled with the dynamic and complex nature of livelihood resilience, we proposed to construct a mass, damping, stiffness, and pressure matrix by referring to the structural dynamics principle of Fang Yiping et al. [36] to consider rural livelihood resilience as a comprehensive attribute [37]. From the causes of resilience [32], we combined key elements and resilience processes to achieve assessment method evolution from dimensionality to structure [38], so as to quantitatively assess livelihood resilience level of rural residents in China over the past four decades.

The objectives of this study include the following four aspects: (1) to establish a livelihood resilience evaluation index system based on a structural dynamics model; (2) to analyze the change trend of the livelihood resilience of rural residents in China between 1980 and 2020; (3) to predict the development trend of livelihood resilience of rural residents in China between 2021 and 2030; and (4) to identify the influencing factors of livelihood resilience of rural residents. The significance of the study is mainly reflected in that, firstly, it enriches the theoretical study on the evaluation method of livelihood resilience of rural residents. Secondly, it predicts the livelihood resilience of Chinese rural residents in the next ten years, which can provide support and reference for the precise implementation of China’s rural revitalization strategy. Finally, it identifies the main constraints to the livelihood resilience of rural residents, which can help reduce the negative impact of natural disasters on the livelihoods of rural residents and enhance their resilience and recover from disasters.

2. Materials and Methods

2.1. Study Area

There are 23 provinces, 5 autonomous regions, 4 municipalities, and 2 special administrative regions in China. With a large geographical area and nearly 700,000 villages, rural areas account for a relatively large proportion of China’s population. In 2020, China’s rural population accounted for 36.11%, and there are still 510 million rural residents and a large agricultural population. According to the statistics, the overall income of rural residents has increased by an average of nearly 10% per year over the past 40 years. Although the overall income of rural residents has gradually increased, and the rural economy has entered a virtuous cycle, according to the statistics, the current economic gap between urban and rural areas is still relatively obvious, and the income ratio between urban and rural residents still reaches 2.56:1, the per capita disposable income of rural residents is also lower than that of national residents. In addition, due to the highly complex geographical and climatic conditions in China, the variety and frequency of natural disasters, coupled with the high level of dependence of rural residents on the natural environment, large-scale land development and agricultural production activities further increase the frequency of various disasters, making the negative impact of natural disasters even more significant. Common natural disasters in China, such as local high temperature and drought, cold waves and early spring crop freezes, seasonal heavy rainfall and flooding, and crop pests and diseases, etc., have a significant impact on reduced agricultural production. A series of consequences caused by natural disasters have become important factors restricting the growth and development of the rural economy. Long-term research on rural residents’ livelihood since China’s reform and opening up can truly show the great changes of rural development in China and explore the reasons for the livelihood resilience changes of rural residents. Therefore, it is appropriate to explore the livelihood resilience of Chinese rural residents in the context of natural disasters.

2.2. Data Sources

The original data of this study are from the published materials such as “China Rural Statistical Yearbook” (1981–2021), “China Statistical Yearbook” (1981–2021) and annual statistical bulletins, and some missing values are filled by interpolation method.

2.3. Methods

2.3.1. Structural Dynamics Model

Structural dynamics focuses on the study of the response of the structure to the load that it bears and describes the bearing capacity and dynamic characteristics of the structure from the dynamic process [39]. In recent years, structural dynamics has gradually become an important tool for the study of complex social-ecological and dynamical systems. Due to its ability to analyze causal relationships and identify the impact of different structural levels and factors on the system, this method has strong advantages in assessing changes in livelihood systems in the context of natural disasters.

Typically, the structural equation is a system of second-order ordinary differential equations with the matrix form [36]:

$$M\stackrel{••}{\chi }(t)+C\stackrel{•}{\chi }(t)+K\chi (t)=f(t)$$

(1)

In the formula, $x\left(t\right)$ is the generalized coordinate vector (displacement), which is a function of time t, and the point on it represents the derivative of time;$M,C$ and $K$ correspond to the mass matrix, damping matrix and stiffness matrix of $x\left(t\right)$ respectively; $f\left(t\right)$ represents the dynamic load vector (pressure).

Multiply both sides of Equation (1) by $M^{-1}$ to obtain Equation (2):

$$\stackrel{••}{x}(t)+M^{-1}C\stackrel{•}{x}(t)+M^{-1}Kx(t)=M^{-1}f(t)$$

(2)

By introducing the state variable $\mathrm{q}={\left[x\stackrel{•}{x}\right]}^T$, the Equation (1) is transformed into:

$$\stackrel{•}{\mathrm{q}}=Dq+p$$

(3)

$$D=\left[\begin{array}{cc}0& I\\ -M^{-1}K& -M^{-1}C\end{array}\right],\mathrm{p}=\left[\begin{array}{c}0\\ M^{-1}f(t)\end{array}\right]$$

(4)

In Equation (4), I represents the identity matrix, and 0 is a zero matrix. Equation (4) is rewritten as:

$$q=e^{Dt}q(0)+{\displaystyle {\int }_0^t{e}^{D(t-\tau )}P(\tau )d\tau }$$

(5)

The displacement $x\left(t\right)$ is calculated by the LSIM function in MATLAB and Equation (3). The magnitude and direction of change are important characteristics of the matrix, and the eigenvalue of the matrix refers to the amplitude of the system under pressure. As can be seen from Equation (1), the right side of the equation is the external pressure that the system is subjected to, the left side is the response state of the system, and the system state changes with the pressure on the right side of the equation. Therefore, the average amplitude is defined as the load capacity of the system, that is, the load capacity of the system is:

$$FC=\lambda (M+C+K)$$

(6)

In the Equation (6), $\lambda$ represents the eigenvalue.

According to the principles of structural dynamics, when external pressure acts on the system, the system will be displaced. The displacement amplitude depends on the three parameters of mass, damping and stiffness. In this paper, the load capacity (FC) per unit displacement is expressed as the livelihood toughness, that is, the livelihood toughness is positively correlated with the load capacity and negatively correlated with the displacement. Under the same load capacity, the smaller the displacement, the stronger the toughness level, and vice versa.

$$LR=\frac{FC}{x(t)}$$

(7)

In the formula, LR is the livelihood resilience index of rural residents, FC is the load capacity of the system under external disaster stress, and $x\left(t\right)$ is the displacement of the system under external pressure or other pressures.

2.3.2. ARIMA Prediction Model

The difference autoregressive moving average model (ARIMA model) is essentially a linear model and is a time series forecasting method proposed by Box and Jenkins [40]. It regards the data sequence formed by the prediction object over time as a random sequence and uses a certain mathematical model to approximately describe the sequence. Once the model is identified, it can predict future values from the past and present values of the time series [41]. In general, ARIMA’s modeling and prediction consists of four parts; (1) data stabilization, where, to eliminate the instability of the original data series, differential processing is performed on the data series; (2) model identification, where ACF and PACF are primarily used to determine the model order p and average order q; (3) model estimation and testing, in which the ARIMA (p, d, q) model is used to fit the data, estimate the parameters of the model and check whether it is stationary; and (4) the obtained parameters from the analysis are used to establish a suitable prediction model.

The ARIMA model is used mainly to analyze time series by specifying three parameters p, d, q, that is, to describe the autoregressive order p, the difference order d and the moving average order q, denoted as ARIMA(p, d, q), which means:

$${\phi }_{\mathrm{p}}{\nabla }^{\mathrm{d}}{Z}_{\mathrm{t}}={\theta }_0+{\theta }_{\mathrm{q}}\left(B\right){\alpha }_{\mathrm{t}}$$

(8)

In the formula, $Z_{\mathrm{t}}$is the original sequence; ${\alpha }_{\mathrm{t}}$ is the white noise sequence, which is a sequence of random variables that have nothing to do with each other, the mean value is 0, and the variance is ${\sigma }^2$; B is the backshift operator, which defines is $B{Z}_t=Z_{t-1}$, so $B^{\mathrm{m}}{Z}_{\mathrm{t}}=Z_{\mathrm{t}-\mathrm{m}}$; ${\phi }_{\mathrm{p}}$ is an autoregressive operator, ${\phi }_{\mathrm{p}}\left(B\right)=1-{\phi }_{1}B-{\phi }_2{B}^2-\dots -{\phi }_{\mathrm{p}}{B}^{\mathrm{p}}$, $\mathrm{p}$ is the moving average order of the model; ${\phi }_{\mathrm{p}}\left(B\right)=1-{\phi }_{1}B-{\phi }_2{B}^2-\dots -{\phi }_{\mathrm{p}}{B}^{\mathrm{p}}$, $\mathrm{p}$ is the moving average order of the model;${\theta }_{\mathrm{q} }$ is the moving average operator, ${\theta }_{\mathrm{q}}\left(B\right)=1-{\theta }_{1}B-{\theta }_2{B}^2-\dots -{\theta }_{\mathrm{q}}{B}^{\mathrm{q}}$, q is the moving average order of the model; $\nabla$ is the backward difference operator, which can be represented by $B$, because, $\nabla Z_{\mathrm{t}}=Z_{\mathrm{t}}-Z_{\mathrm{t}-1}=\left(1-B\right)Z_{\mathrm{t}}$; θ₀ is a constant term, defining ${\theta }_0=\mathrm{u}\left(1-{\phi }_1-{\phi }_2-\dots -{\phi }_{\mathrm{p}}\right)$, where $\mathrm{u}$ is the mean.

2.3.3. Ridge Regression Analysis

Most of the existing empirical methods on the influencing factors of livelihood resilience are based on the multiple linear regression analysis of ordinary least squares (OLS), but the premise of applying this regression method is that the independent variables must be independent of each other [42]. It is a complex and multivariable system, and this paper considers the influencing factors of livelihood resilience in the context of natural disasters and based on long-term series, and there may be strong correlations between the influencing factors and variables, so this condition is used in the analysis of livelihood resilience in this paper. It is difficult to meet the toughness factor. Ridge regression analysis is a modified least squares estimation method, and it is also a regression method to solve such problems. When the independent variables have multicollinearity problems, this method can provide a more stable estimation than the least squares method, and the regression coefficients of the standard deviation are also smaller than those estimated by the least squares method. Therefore, to weaken the effect of multi-collinearity, this paper uses a ridge regression model to analyze the influencing factors of rural residents’ livelihood resilience.

Ridge regression estimation is defined as [42]:

$$\stackrel{\Lambda }{\beta }\left(\mathrm{k}\right)={\left(X^{T}X+kI\right)}^{-1}{X}^{T}Y,0<k<+\infty$$

(9)

The general equation of ridge regression can be obtained by transforming the equation form:

$$\left(X^{T}X+kI\right)\stackrel{\Lambda }{\beta }\left(\mathrm{k}\right)=X^{T}Y$$

(10)

In the formula, $X$ represents the $\mathrm{m}\times n$ order coefficient matrix, which consists of standardized × variables; $\stackrel{\Lambda }{\beta }\left(\mathrm{k}\right)$ represents the estimated coefficient vector; $I$ represents the identity matrix; $k$ represents the ridge regression coefficients; appropriate coefficients are obtained by ridge regression analysis.

2.4. Construction of Evaluation Index System

According to the characteristics of structural dynamics, the level of resilience of rural residents’ livelihoods is a comprehensive reflection of the state of mass, damping, stiffness, and pressure. This paper draws on relevant research results [36,43], combined with a resilience analysis framework and the livelihood characteristics of rural residents in China. The four dimensions of livelihood characteristics—livelihood quality, livelihood promotion, livelihood supply, and disaster stress—construct an evaluation index system for rural residents’ livelihood resilience (

Table 1. The livelihood resilience evaluation index system of rural residents.

Dimension	Indicator	Unit	Quality	Matrix	Connotation
Livelihood quality	Engel’s coefficient	%	Negative	Mass matrix	Reflects the living conditions and well-being of rural residents, and emphasizes the quality of life and development level.
	Rural per capita disposable income	Yuan	Positive
	Rural per capita education, culture and entertainment consumption expenditures	Yuan	Positive
	Rural residents’ per capita medical consumption expenditures	Yuan	Positive
	Beds in medical and health institutions for 10,000 people	-	Positive
Livelihood promotion	The proportion of local government’s per capita finacial income and expenditure	%	Negative	Damping matrix	Refers to the potential to restore livelihoods under the stress of disasters, and emphasizes the ability of a population to sustain livelihoods under the influence of the government and the family.
	Rural per capita investment in fixed assets	Yuan	Positive
	Rural per capita minimum living security	Yuan	Positive
	Per capita agricultural, forestry and water fiscal expenditure	Yuan	Positive
	Per capita natural science and technology research expenditure	Yuan	Positive
Livelihood supply	Per capita arable land area	Mu	Positive	Stiffness matrix	Refers to basic life needs, including food, health and basic security of life, and is a direct feature of livelihood insecurity.
	Rural per capita grain output	Ton	Positive
	Rural per capita agricultural machinery total power	Watt	Positive
	Rural per capita housing construction area	m2	Positive
	Rural per capita electricity consumption	KWH	Positive
Disaster stress	The proportion of rainstorm and flood-affected area in total planting area	%	Negative	Stress matrix	Shows the disaster stress experienced by rural residents.
	The proportion of drought-affected area in total planting area	%	Negative
	The proportion of wind and hail disaster-affected area in total planting area	%	Negative
	The proportion of low-temperature disaster-affected area in total planting area	%	Negative

Note: The attribute of the indicators could be divided into positive and negative. The larger the value of the positive indicator, the better the evaluation result, while the larger the value of the negative indicator, the worse the evaluation result.

).

The livelihood quality dimension corresponds to the mass matrix. Mass, in structural dynamics, is a fundamental property of matter and exists in all physical systems [39], meaning that this concept is applicable to livelihood systems and can be defined as a livelihood quality. Livelihood quality reflects the livelihood status and well-being level of rural residents, emphasizes their quality of life and development level, and reflects the endowment and quality of their livelihood resources. Engel’s coefficient, rural per capita disposable income, rural per capita education, culture and entertainment consumption expenditures, rural residents’ per capita medical consumption expenditures, and beds in medical and health institutions per 10,000 people were selected as indicators of livelihood quality.

The livelihood promotion dimension corresponds to the damping matrix. In structural dynamics, damping is an equivalent physical quantity. Different damping forms are used to simulate real physical processes for different problems. This can be said to be a compromise, allowing for the application of this concept in livelihood systems and allowing it to be defined as a livelihood promotion. Livelihood promotion is the livelihood recovery potential of rural residents under the threat of natural disasters, emphasizing the ability to maintain livelihoods under the influence of governments and households [44]. The proportion of local government’s per capita financial income and expenditure, rural per capita investment in fixed assets, rural per capita minimum living security, per capita agricultural, forestry and water fiscal expenditure, and per capita natural science and technology research expenditure were selected as livelihood promotion indicators.

The livelihood supply dimension corresponds to the stiffness matrix. In structural dynamics, the stiffness structure is more rigid, which reduces dynamic effects and makes it more dependent on static forces and displacements [39]. Although their objects are different, dynamic systems are similar to social and subsistence systems and applying this concept to subsistence systems defines it as a subsistence provision. The livelihood supply mainly refers to the supply level of rural residents’ livelihood, including the normal maintenance of rural food, population health and basic living conditions, as well as the security level of a basic living guarantee for rural residents to cope with emergencies. The per capita arable land area, the rural per capita grain output, the rural per capita agricultural machinery total power, the rural per capita housing construction area, and the rural per capita electricity consumption were selected as the indicators of livelihood supply.

The disaster stress dimension corresponds to the stress matrix, and livelihoods and the environment have always had a crucial connection. The daily livelihoods of rural residents are extremely vulnerable to multiple shocks and disturbances caused by climate change, and these disturbances have adversely affected their livelihood systems. Natural disasters have also become important factors hindering the sustainable livelihoods of rural residents. Disaster stress reflects the state of disaster experienced by rural residents. According to the actual situation of China’s rural areas, the proportion of rainstorm and flood-affected area in total planting area, the proportion of drought-affected area in total planting area, the proportion of wind and hail disaster-affected area in total planting area, and the proportion of low-temperature disaster-affected area in total planting area were selected as indicators of disaster stress.

2.5. Influencing Factors of Livelihood Resilience of Rural Residents

To further explore the impact of factors at different dimensions on the livelihood resilience of rural residents, and with reference to relevant literature [23,36], we took livelihood resilience as a dependent variable, and eight indicators from four dimensions of economic foundation, science and education level, social characteristics and disaster resistance as independent variables (

Table 2. Factors affecting livelihood resilience.

Dimension	Variable	Code	Explanation
Economic foundation	Rural per capita consumption level	X1	The more rural per capita GDP, or the higher per capita consumption level, the better the economic level.
	Rural per capita GDP	X2
Science and education level	The number of health technicians per thousand population	X3	The higher the number of health technicians per thousand people, or the wider the reach of higher education, the higher the level of science and education in the region.
	Higher education penetration level	X4
Social characteristics	Proportion of people employed in primary industries	X5	The lower the proportion of people employed in primary industries, or the higher the penetration of mobile phones, the more diverse the means of living, and therefore the more stable the social system..
	Mobile phone penetration rate	X6
Disaster resistance	Effective irrigation area	X7	The larger the area of effective irrigation or the larger the area of embankment protected cultivated land, the higher the level of disaster resistance infrastructure construction.
	Area of embankment-protected arable land	X8

), in order to conduct ridge regression analysis. Among the independent variables, economic foundation includes rural per capita consumption level (X1) and rural per capita GDP (X2), science and education level includes number of health technicians per thousand population (X3) and higher education level (X4), social characteristics include proportion of people employed in primary industries (X5) and mobile phone penetration rate (X6), and finally, disaster resistance includes effective irrigation area (X7) and embankment protected cultivated land (X8). After data standardization, SPSSAU was used to perform ridge regression analysis on the standardization results of various influencing factors and the livelihood resilience index from 1980 to 2020.

2.6. Data Processing

The key steps in data processing consist of the following:

Step 1: Data normalization. The extreme value method was used to standardize the original data of each index to eliminate the influence of the dimension difference between variables. As one of the most accurate free scaling techniques, the current index standardization method has been widely used in previous studies [45,46]. Equation (11) and Equation (12) were applied for the normalization of positive and negative indicators, respectively. In the formula, $X_{i\max }$ and $X_{i\min }$ are the maximum and minimum values of the i-th index, respectively.

Step 2: Solve the structural dynamics equation. The LISM function of MATLAB software predicts the trajectory of a particular system described by the projection in state-space form [47]. According to the mass, damping, stiffness and pressure matrices in Equation (5), the displacement variable $\left(x\left(t\right)\right)$ was calculated using the LISM function in MATLAB software.

Step 3: Measure the livelihood resilience level of rural residents and compare and analyze the characteristics of four-dimensional structural changes. According to the displacement obtained by Equation (2) and the load force of the system obtained by Equation (6), the livelihood resilience level LR of rural residents was determined by Equation (7), and the four-dimensional structure of each year can be obtained numerical value.

Step 4: Predict the changing trend of rural residents’ livelihood resilience. Using the results obtained above, the appropriate parameters p, d, q in the ARIMA (p, d, q) model were determined through the time series scatter plot, ADF test and partial autocorrelation plot, and the model was used to predict.

Step 5: Analyze the impact of factors at different levels on the level of livelihood resilience. Taking the level of livelihood resilience as the dependent variable and selecting 8 indicators from the four levels of economic foundation, science and education level, social characteristics, and disaster resistance as independent variables, a ridge regression analysis was carried out.

$${{\displaystyle X}}_i^{\text{'}}=\frac{X_i-X_{i\min }}{X_{i\max }-X_{i\min }}$$

(11)

$${{\displaystyle X}}_i^{\text{'}}=\frac{X_{i\max }-X_i}{X_{i\max }-X_{i\min }}$$

(12)

3. Results

3.1. Evaluation Results of Livelihood Resilience of Rural Residents in China

3.1.1. The Changing Trend of Livelihood Resilience of Rural Residents in China

Since 1980, the livelihood resilience of rural residents in China has fluctuated greatly, but the overall trend is upward (Figure 1). During the period between 1980 and 1984, there was a large-scale drought across the country, and social and economic construction was still in the initial stage of reform and opening up. The infrastructure construction in rural areas was still incomplete, resulting in major losses due to disaster. Therefore, the livelihood resilience index of rural residents showed a significant downward trend. From 1985 to 2002, the rural livelihood resilience index showed a trend of up and down fluctuations, and the trend of increase and decrease of the curve was not obvious. Among these, the livelihood resilience index showed a downward trend in 1988 and 1996, and the trough of the fluctuation curve appeared in 1991 and 1999. This was mainly attributed to the occurrence of global climatic anomalies during this period, and the frequent occurrence of meteorological disasters in China, especially the widespread occurrence of heavy rainfall and flooding, which particularly affected rural areas. From 2003 to 2012, the overall livelihood resilience of rural residents showed rapid growth, however, during 2013 to 2015, the country was affected by several extreme weather and climate events, such as strong typhoons causing multiple floods, and the livelihood resilience index took a significant downward turn. From 2016 to 2020, the livelihood resilience index showed a high and rapid upward trend.

3.1.2. The Changing Characteristics of The Four-Dimensional Structure of Livelihood Resilience of Rural Residents in China

Comparing the evaluation values of various dimensions of livelihood resilience of rural residents in China (

Table 3. The evaluation values of various dimensions of livelihood resilience.

Dimension	F-Value	t-Value	Contribution Rate
Livelihood quality	45.562 ***	4.694 ***	35.2%
Livelihood promotion		1.953 ***	24.4%
Livelihood provision		0.394 ***	22%
Disaster stress		−5.331 ***	60.1%

Note: *** indicate significant at the level of 1%.

), it can be seen that the level of rural residents’ livelihood resilience is significantly positively correlated with livelihood quality, livelihood promotion and livelihood supply, and significantly negatively correlated with disaster stress. The contribution rate of the four structural dimensions are, in descending order, disaster stress, livelihood quality, livelihood promotion, and livelihood supply. Among these, the contribution rate of disaster stress is significantly higher than that of other structural dimensions, which is the dominant factor affecting the overall level of livelihood resilience; the contribution rates of livelihood quality, livelihood promotion, and livelihood supply are relatively similar, and they have become the main forces jointly promoting the overall level of livelihood resilience.

The trend in the four-dimensional structure of livelihood resilience shows an overall increasing trend in the curves of livelihood quality, livelihood promotion and livelihood supply (Figure 2). In particular, there was a rapid growth trend after 2000, reflecting the level of social-economic development and infrastructure construction of rural China in the 21st century. In addition, disaster stress also showed a fluctuating growth trend, especially at the time point when the livelihood resilience declined and then increased rapidly. This shows that the level of livelihood resilience in China was significantly affected by natural disasters, and the disaster stress did not change significantly due to the improvement of livelihood quality, livelihood promotion and livelihood supply.

From 1980 to 2020, the overall livelihood quality index of rural residents in China showed a rapid upward trend. At the indicator level, the improvement of livelihood quality was mainly due to the decline of the Engel coefficient, the increase of access to medical care, the improvement of rural per capita education level and the increase of rural per capita disposable income. China’s Engel coefficient declined from 61.8 in 1980 at the beginning of China’s reform and opening-up to 32.7 in 2020, while the number of beds per 10,000 people in medical and health institutions increased from 22.1 to 64.5, and rural per capita medical consumption expenditure also increased significantly. Since the 1980s, China has implemented the household contract responsibility system, and rural production have been significantly improved. At the same time, the recovery of sideline businesses, diversification of operations and the transfer of labor to the secondary and tertiary industries has improved the livelihoods of rural residents. The per capita disposable income in rural areas increased from 1980, and 191.3 yuan rose to 17,131.5 yuan in 2020. In addition, the per capita education expenditure of rural households increased significantly, giving rural residents more opportunities to acquire knowledge and skills, thereby improving their ability to prevent and resist livelihood risks, resulting in an effective improvement in quality of life.

From 1980 to 2020, the overall livelihood promotion index showed a steady upward trend. From 1980 to 1994, the growth of the livelihood promotion index was relatively slow. During this period, the reduction of rural per capita minimum living security expenditure had a certain negative impact on the improvement of rural residents’ livelihood promotion. However, since 1995, the rising speed of the livelihood promotion index increased significantly. During this period, the proportion of local governments’ per capita financial income and expenditure gradually increased, the local government’s financial supply capacity steadily improved and the intensity of rural residents’ investment in fixed assets further strengthened. Rural per capita minimum living security increased from 0.23 yuan in 1995 to 279.78 yuan in 2020. In years of serious disasters, the government’s financial assistance can enable rural residents to return to normal production and life as soon as possible. With investment in technology, the large amount of rural capital promoted the renewal and improvement of agricultural production technology and the significant improvement of rural production efficiency, especially after the Ministry of Agriculture and Rural Development released the Technical Guidelines for Green Development of Agriculture (2018–2030) in 2018. The construction of rural and agricultural modernization has since received wider attention. It has brought about the effective use of rural labor resources, gradually balancing the relationship between capital and labor resources, providing more security for the livelihood of rural residents, while driving the development of secondary and tertiary industries and providing more employment opportunities. It has also prompted society to invest more in rural areas and promote rural economic development, which plays an important role in promoting the stability of the livelihood system of rural residents.

Between 1980 and 2020, the livelihood supply index showed a fluctuating upward trend. During this period, due to adjustments in agricultural structure, ecological farmland, and construction occupation, a large amount of cultivated land was lost in China, and the per capita cultivated land area decreased from 1.51 mu to 1.17 mu between 1980 and 1995. After 1996, per capita arable land increased through agricultural structural adjustment, and then gradually decreased. In general, due to the continuous changes in agricultural structural adjustment in the “Government Work Reports” each year, the per capita arable land was also constantly changing. Meanwhile, due to uncertain factors such as natural disasters and the continuous adjustment of arable land over the past 41 years, the per capita grain output in rural areas has fluctuated irregularly, resulting in fluctuations in the livelihood supply index. Rural per capita agricultural machinery total power continued to grow between 1980 and 2020, which helped to improve agricultural production conditions, while reducing, to an extent, the losses caused to agricultural production by natural disasters and contributing to the growth of rural residents’ income. Rural per capita housing construction area and per capita electricity consumption in rural areas also increased year on year, reflecting improvements in the rural productivity and living standards of rural residents, thus creating a virtuous circle. This provided greater development impetus for production and life of rural residents and has had a positive impact on rural residents’ livelihood supply.

Between 1980 and 2020, the disaster stress index changed periodically. Since 1980, extreme weather has become frequent in China, resulting in a frequency of natural disasters, especially meteorological disasters. The peaks and troughs of the disaster stress index curve for rural residents appeared alternately, and each rapid increase in disaster stress represented a major natural disaster. From 1980 to 1983, a rare major drought occurred in China, followed by major floods in 1991 and 1998, and a large-scale drought in 2000, the affected area of which reached a magnitude not seen in 41 years, accounting for 25% of the total planting area. Since 2010, disaster stress has shown a fluctuating and rising trend, and has remained at a relatively high level. This was mainly due to the increased frequency of extreme disastrous weather in the context of climate change. In addition to the direct economic losses caused by natural disasters, rural residents have had to invest more money in disaster response, and this additional investment indirectly affected their income, which in turn has caused significant social-economic losses to rural China.

3.2. Prediction of the Changing Trend of Rural Residents’ Livelihood Resilience

Based on the structural dynamics model, the ARIMA model was established according to the constructed livelihood resilience evaluation model. Due to the uncertainty and unpredictability of disaster stress, only livelihood quality, livelihood promotion, livelihood supply, and livelihood resilience were selected for trend prediction.

The results of the forecast in Figure 3 show a slow increasing trend of livelihood resilience for rural residents in China from 2021 to 2030. By 2030, the livelihood resilience index will have increased from 1.200 in 2020 to 1.520, with an insignificant overall trend, and the growth rate of livelihood resilience will be lower than that of the early 21st century, with a small decline in 2023–2024. This indicates that the future growth of rural residents’ livelihood resilience will slow according to the current developmental conditions, and there is a possibility of a downward trend in livelihood resilience. It can be seen that with the improvement of their production and living conditions and the support and assistance of the government, the volatility of the livelihood system of rural residents will be reduced, and the livelihood of rural residents will enter a stage of stable development in the future. However, no matter how the external conditions are improved, uncertain factors such as natural disasters will always exist and impact the overall livelihood resilience level, so that rural residents’ livelihood resilience will still fluctuate slightly. Therefore, in the future, it is necessary to pay attention to the stability and sustainability of rural residents’ livelihood resilience.

The livelihood quality forecast from 2021 to 2030 shows a steady upward trend, which is in line with the 14th Five-Year Plan (2021–2025) for China’s national economic and social development. The 14th Five-Year Plan will further improve people’s livelihood and well-being, consolidate and expand the achievements of poverty alleviation, comprehensively promote the rural revitalization strategy, and promote the construction of public health emergency support and rural education support. In addition, the predicted trend of livelihood quality also reflects that the long-term impact of the new epidemic on the livelihood of rural residents will be weaker, and the livelihood potential of rural residents will be further released. It can be seen that improving the quality of livelihoods can lead to a more robust development of livelihood resilience among rural residents. The predicted results for livelihood supply show the same slow growth trend, and the growth rate is similar to the quality of livelihoods. It can be expected that as rural society develops, basic livelihood security such as food, clothing, shelter and transport will gradually stabilize, productive living space will be optimized, the human living environment will be improved, and when basic livelihood needs are met, the focus will shift to other priority areas in the future.

It is predicted that livelihood promotion will grow rapidly from 2021 to 2030, with the highest growth rate level since the 21st century, indicating that the future strategic development of rural areas in China will remain a key focus of national construction and development. The national government will continue to increase investment in rural construction, improve the multi-level social security system, and speed up the filling of shortcomings in areas such as infrastructure, agriculture and rural areas, disaster prevention and mitigation, and industrial structure. It can be seen that strengthening the guarantee of rural development factors can promote livelihoods, and the improvement of livelihood promotion can effectively prevent the shock to rural residents caused by unexpected events. Therefore, livelihood promotion will be important in the implementation of the rural revitalization strategy in the coming decade, which will also help to incorporate the fundamental role of rural economic formats in social development.

It is predicted that livelihood supply will show a slow growth trend from 2021 to 2030, with its growth rate close to that of livelihood quality. With the process of industrialization, urbanization and aging, the per capita arable land area has, since 2018, been lower than the average of the previous 41 years. As one of the foundations of livelihood capital, the protection of arable land in the future will be a challenge for all rural areas, and will also have a significant impact on the overall development trend of the livelihood supply. Therefore, the 14th Five-Year Plan in China gives priority to the development of agriculture, adheres to the strictest cultivated land protection system, and improves the quantity and quality of cultivated land, so as to provide basic support for the security of livelihood supply in the future. In addition, with the development of rural society, basic living security such as food, clothing, housing and transportation will be gradually stabilized, production and living space will be optimized, and the living environment will be improved. When these basic living needs are met, the future focus will be shifted to other areas, thus making the trend of livelihood supply slow down.

3.3. Ridge Regression Results of Influencing Factors of Livelihood Resilience

According to the regression results (

Table 4. Ridge regression results of influencing factors of livelihood resilience.

Dimension	Variable	Unstandardized Coefficient		Standardized Coefficient	t-Value	p-Value	R2	F-Value
		B	SE
Constant	C	0.301	0.05	-	5.999	0.000 ***	0.795	15.540 ***
Economic foundation	X1	0.165	0.071	0.135	2.31	0.028 **
	X2	0.348	0.062	0.293	5.634	0.000 ***
Science and education level	X3	0.208	0.07	0.16	2.996	0.005 ***
	X4	0.2	0.073	0.205	2.761	0.009 ***
Social characteristics	X5	−0.118	0.106	−0.105	−1.116	0.273
	X6	0.196	0.044	0.318	4.462	0.000 ***
Disaster resistance	X7	0.016	0.086	0.015	0.184	0.855
	X8	0.281	0.093	0.271	3.025	0.005 ***

Note: *** and ** respectively indicate significant at the level of 1% and 5%.

), the R2 of the model is 0.795, indicating that 8 variables could explain 79.5% of the changes in livelihood resilience. And the model passed the F test (F = 15.540, p < 0.05), indicating that at least one variable would have an impact on livelihood resilience.

In the economic foundation dimension, the coefficient of rural per capita consumption level (X1) and rural per capita GDP (X2) are 0.165 (p < 0.05) and 0.348 (p < 0.05) respectively, indicating that rural per capita consumption level and rural per capita GDP have a significant positive impact on livelihood resilience.

In the dimension of education level, the regression coefficients of the number of health technicians per thousand population (X3) and higher education penetration level (X4) are 0.208 (p < 0.05) and 0.2 (p < 0.05) respectively, indicating that the higher the value of these two variables, the stronger the sustainable development ability of rural residents’ livelihood system. The number of health technicians per thousand people reflects the importance the state attaches to health undertakings.

In terms of the social characteristics dimension, the proportion of people employed in primary industries (X5) do not pass the significance test (p > 0.1), while the mobile phone penetration rate (X6) is 0.196 (p < 0.05), indicating that mobile phone penetration has a significant positive impact on livelihood resilience.

In terms of the disaster resistance dimension, effective irrigated area (X7) does not pass the significance test (p >0.1), while the regression coefficient of area of embankment-protected arable land (X8) is 0.281 (p < 0.05), indicating that the increase of embankment protected cultivated land has a significant positive impact on livelihood resilience.

4. Discussion

4.1. Evaluation and Prediction of Livelihood Resilience

The comprehensive evaluation results show that the livelihood resilience of rural residents in China fluctuated between 1980 and 2020, but generally showed an upward trend. During this period, meteorological disasters occurred frequently in China due to the abnormal global climate [48]. The four major fluctuations of livelihood resilience were all caused by natural disasters, such as droughts or floods in 1983, 1991 and 1998 and Typhoon Rammasun in 2014 [49,50], which made livelihood resilience turn down to the local ‘bottom’. From 2016 to 2020, the livelihoods resilience index showed a rapid upward trend with no major disasters and an improvement in rural socio-economic conditions.

By examining the evaluation value of the four dimensions of resilience, it can be found that the livelihood resilience of rural residents is significantly positively related to livelihood quality, livelihood promotion and livelihood supply, and significantly negatively correlated with disaster stress. Meanwhile, the contribution rates of the four dimensions are disaster stress, livelihood quality, livelihood promotion and livelihood supply. It is worth noting that the contribution rate of disaster stress to the change in livelihood resilience is significantly higher than that of the other three dimensions, and therefore it can be regarded as the main factor influencing the overall level of livelihood resilience. According to the trend of the four dimensions between 1980 and 2020, the livelihood quality, livelihood promotion and livelihood supply curves showed relatively stable upward trends, while the disaster stress curve had the characteristic of large cyclical fluctuation but still showed an upward trend. Livelihood quality, specifically, showed a rapid upward trend. This is mainly due to the implementation of the household contract responsibility system in 1980s and the transfer of rural labor to non-agricultural industry [51], which has greatly improved the life-quality of rural residents. In addition, the increasing emphasis on education among rural residents has greatly improved their ability to withstand life risks [52]. The livelihood promotion showed a steady increasing trend, in which the government has played a powerful role as a driving force. During the study period, the government’s financial supply capacity has steadily improved, and the expenditure on rural residents’ livelihood security and disaster resistance has also been greatly increased, as well as the investment in agricultural science and technology [53], which has provided a solid guarantee for livelihood resilience. The livelihood supply index curve showed an upward, if slightly fluctuating, trend. As a result of annual agricultural restructuring, the area of cropland is in a constant state of change [54], thus making livelihood supply unstable. However, the modernization of agriculture and the improvement of rural housing has provided a strong developmental impetus to rural regions [55], which in turn has had a strong positive impact on the improvement of livelihood supply. In the context of climate change, natural disasters occur periodically, causing huge economic losses to rural societies in China [56].

Based on the prediction of livelihood resilience from 2021 to 2030, it can be seen that livelihood promotion will increase rapidly, while livelihood supply and livelihood quality will increase slowly but steadily. Further, the overall livelihood resilience will grow very slowly, slower than livelihood promotion, livelihood supply and livelihood quality. It can be seen that, with the improvement of rural production and living conditions and the role of government assistance [57], the volatility of rural residents’ livelihood system will decrease in the next ten years, and livelihood resilience will enter a stage of stable development.

4.2. Influencing Factors of Rural Residents’ Livelihood Resilience

In the economic foundation dimension, the per capita consumption in rural areas rose rapidly from 178 yuan in 1980 to 16,063 yuan in 2020, which shows that the policies for rural revitalization and poverty alleviation have been effective and have significantly improved the quality of life of rural residents. between 1980 and 2020, rural per capita GDP showed a trend of annual growth. Since 2000 in particular, rural per capita GDP has grown rapidly, this is because at the turn of the century, the state began to transfer payments to rural areas on a large scale, providing basic security for rural residents and alleviating absolute poverty in rural areas [58]. In addition, the abolition of agricultural tax in 2006 [59], the promotion of modern agriculture in 2007 and the construction of agricultural infrastructure in 2008 have further promoted the construction of new countryside [60], resulting in a significant increase in rural GDP and farmers’ income. As a result, the livelihood resilience of China’s rural residents has steadily increased.

In the dimension of education level, the number of health technicians per thousand people reflects the importance the state attaches to health undertakings. The growth rate of health technicians per thousand people had been relatively slow from 1980 to 2000, but showed a trend of rapid growth after 2001, something which is mainly attributed to the reform of national health undertakings [61]. For instance, the new rural cooperative medical system was piloted in 2003 [62], the reform of the medical system was comprehensively deepened in 2009 [61], and the National Health Commission was established in 2018. Higher education penetration level can measure regional education level to some extent. From 1980 to 1998, higher education penetration level was relatively low. However, in 1999, the Ministry of Education launched the Education Revitalization Action Plan for the 21st Century, expanding the ordinary university enrollment [63]. Since then, the proportion of college students has increased year by year. With the joint efforts of the state and of rural residents, the education level of the rural labor force has been continuously improved and rural residents have gained more livelihood capacity through learning, which has a positive impact on the resilience level of livelihood.

In terms of the social characteristics dimension, in 1980, the mobile phone penetration rate was only 0.43%. And by 2012, this rate has exceeded 100%. In 2020, this rate has reached 125.8%. The 1990s saw the fastest growth in mobile phone penetration [64,65]. In a word, the popularity of mobile phones facilitates communication among residents, especially in rural areas where information is relatively isolated. It enables the rapid circulation of livelihood capital of rural residents [66], thus enriching livelihood strategies and improving the livelihood resilience of rural residents.

In terms of the disaster resistance dimension, China’s climate characteristics determine that without the protection of irrigation and water conservancy facilities, there will be no stable agricultural yield [67]. Between 1980 and 2020, while the per capita cultivated land area showed a trend of fluctuating decrease, the area of embankment protected cultivated land still showed a trend of fluctuating increase. After the 1990s in particular, China implemented the eighth five-year Plan (1991–1995), which made clear arrangements for the development of agriculture and water conservancy, and greatly promoted the infrastructure construction [68]. During this period, embankment-protected cultivated land was increased greatly, which improved the disaster resistance level of agriculture and the resilience level of rural residents’ livelihoods.

4.3. Policy Suggestions

Based on the evaluation and prediction of livelihood resilience and the analysis of influencing factors, four policy suggestions are provided for the subsequent improvement of rural residents’ livelihood resilience.

Firstly, the rural revitalization strategy should be taken as an opportunity to promote the optimization and upgrading of economic structure [69,70]. Although the government is taking advantage of the rural revitalization strategy to vigorously support the secondary and tertiary industries in rural China, there are still relatively few employment opportunities there due to the problems of having a single industrial structure, a short tourism industry chain and a great dependence on the natural environment [71]. Therefore, it is necessary to promote the integrated development of primary, secondary and tertiary industries, and realize the dynamic development pattern of government support, industrial integration and rural residents’ participation [72].

Secondly, government should increase the support for rural technical personnel and investment in higher education [73]. At present, education investment in rural areas has been increased, but there is still a problem of uneven distribution of education resources between urban and rural areas [74]. Therefore, more high-quality education resources should be allocated in rural areas, while ensuring that rural students can get equal opportunities to receive education. In addition, local governments should optimize the allocation of human resources while encouraging college students to return to their hometowns to start businesses and find jobs [75], and steadily improve the livelihood resilience in rural areas.

Thirdly, there should be a popularization of knowledge of climate change and natural disasters, and a strengthening of disaster prevention guidance for rural residents [76]. Natural disasters caused by climate change have a great impact on agriculture. Natural disasters such as typhoons, rainstorms, sandstorms, and cold waves often cause a large area of crop yield reduction or even failure [77]. As the main agricultural body, rural residents can effectively reduce disaster losses if they can master the prevention and response measures of climate change and natural disasters and make rational use of agricultural biotechnology and various ecological maintenance means.

Fourthly, it is necessary to promote rural infrastructure projects and implement rural environmental warning and evaluation mechanisms. This study shows that disaster resistance is an important factor affecting the resilience of rural residents. Irrigation and water conservancy is the lifeblood of agricultural development [78]. Over the years, China has suffered serious losses from flood disasters, and the increase of embankment-protected arable land can directly and effectively alleviate the negative effects of flood and soil erosion on agriculture. In addition, rural environmental warning and evaluation mechanisms must be established [79], which requires promoting the development of agricultural modernization, strengthening the support of rural agricultural technology and equipment, and increasing subsidies for the purchase of agricultural machinery, so as to improve the level of agricultural mechanization and informatization. In this way, in the case of external environmental disturbance and climate change, the corresponding departments can immediately make positive responses and take effective measures to minimize the negative impact of natural disasters on rural areas, so as to realize the symbiotic and harmonious development of rural residents and the environment.

4.4. Research Deficiencies and Prospects

Most existing studies evaluate livelihood resilience based on the framework of livelihood resilience analysis, while this study introduces a structural dynamics model. The model is a structural analysis that includes structural behavior and has strengths in dynamic description, causal analysis and livelihood response estimation, and can play an important role in identifying the structural effects of different factors and realizing the study of dynamic changes in livelihood resilience in the context of natural disasters. It provides a new perspective on the adjustment of rural livelihood systems and response to external disturbances, which is important for reducing disaster losses and scientifically promoting sustainable rural development.

It should be emphasized that there is still a degree of uncertainty in both the dynamic analysis of the past and the prediction of the future trend in livelihood resilience. This study used yearbook data as statistical information, and the research scale was the national scale at the macro level. The research object was only rural resident groups, and there is a need to expand the research on the livelihood resilience of specific groups under the influence of other factors such as policy changes. There is still a need for further research on whether the research method of building a system of resilience indicators through structural dynamics will become a common comprehensive assessment model in the future. In addition, the livelihood resilience index is a comprehensive evaluation index influenced by multiple factors, and there are also some unobservable variables, such as residents’ happiness and behavioral characteristics. Therefore, a structural equation model of livelihood influencing factors can be established to measure potential variables with the help of advanced research tools, while exploring the mechanism of synergy between influencing factors. This would provide theoretical bases for finding the best balance between the construction and management of multiple rural areas at a certain time point and for developing strategies for different behavioral objectives to sustain their livelihood levels.

5. Conclusions

This study used a structural dynamics model to measure the livelihood resilience of Chinese rural residents between 1980 and 2020. The results show that the livelihood resilience of Chinese rural residents fluctuates greatly but shows an overall increasing trend. The livelihood resilience of rural residents is significantly and positively correlated with livelihood quality, livelihood promotion, and livelihood provision, while it is significantly and negatively correlated with disaster stress. In addition, the forecast of the livelihood resilience from 2021 to 2030 shows that the overall increase of livelihood resilience will be slow, and that there is a possibility of it moving downward. Finally, the six variables of rural per capita consumption level, rural per capita GDP, number of health technicians per thousand people, higher education penetration level, mobile phone penetration rate and area of embankment-protected arable land, all have a significant positive impact on livelihood resilience, providing a basis for the subsequent enhancement of livelihood resilience.