Impacts of Eco-Poverty Alleviation Policies on Farmer Livelihood Changes and Response Mechanisms in a Karst Area of China from a Sustainable Perspective

Abstract

Eco-poverty alleviation policies have significant impacts on the changes in farmer household livelihoods. This study developed a multi-dimensional index system, which applies the social cognitive theory and farmer household livelihood capital to analyze the effects of eco-poverty alleviation policies on farmer household livelihood change in a karst area in China. The multivariate logistic, entropy weight, and Technique for Order of Preference by Similarity to Ideal Solution models were used to analyze the responses of 892 farmer households from eight villages in Guizhou Province, China. The results show that the Poverty Alleviation Resettlement Project (PARP) had the highest impact as it enables higher engagement of farmer households in non-agricultural activities, resulting in significant livelihood changes. Among the eco-poverty alleviation policies studied, changes in livelihoods of farmer households are highest from PARP, followed by the Ecological Forest Ranger Project (EFRP), Grain for Green Program (GGP), Forest Ecosystem Compensation Program (FECP), and Single Carbon Sink Program (SCSP). Specifically, GGP received the highest response from farmer households working out-of-province, whereas SCSP received the lowest. EFRP received the highest response from farmer households working in the village. Farmer households in different regions were found to respond differently to various eco-poverty alleviation policies, based on how specific policies can address their practical problems. It is also related to the delayed effects of these policies on their livelihoods. This study provides a theoretical basis for optimizing livelihood improvements for farmers at the regional level, which can aid in formulating strategies in the future to alleviate poverty and revitalize local rural communities.

1. Introduction

Presently, the livelihood of farmer households is one of the most extensively studied subjects in social sciences [1]. Farmer households tend to adjust their livelihoods according to their own interests and risk profiles with the goal of minimizing vulnerabilities to any disruptions [2,3]. They consider the most appropriate livelihood for the benefit of their individual and family development. In addition, farmer households’ resource endowment, livelihood capital, and national and local policies are also considered [4,5,6]. Their livelihood strategies were adjusted according to the issue and implementation of policies [7]. Consequently, their livelihood decisions also affect eco-poverty alleviation policies and their implementation, which could directly or indirectly contribute to policy modification, adjustment, or even termination [8].

With the launch of Targeted Poverty Alleviation Projects in China, ecological and environmental governance has played a role in the success of these projects, with several government departments issuing related ecological policies and measures [9]. The Grain for Green Program (GGP), one of the largest ecological construction programs in the world, aims to protect and improve the ecological environment [10]. The Poverty Alleviation Resettlement Project (PARP) in Chinese villages, implemented through the democratic structures of the Organic Law of Villages Committees and Assemblies, includes all aspects of the relocation decision and the process of rebuilding homes, as well as the requirements for democratic consultation with households, and is undertaken voluntarily by villages. The Ecological Forest Ranger Project (EFRP) is a project designed to protect the ecology and forest resources in China. It uses the project funds of ecological compensation and ecological protection to subsidize some of the local poverty registration households who have the ability to work as ecological forest rangers. The Single Carbon Sink Program (SCSP) is an ecological poverty alleviation project in Guizhou that transforms the ecological value of the forest into economic value. Some poverty registration households meet the standard of carbon-sink trading on the platform. The Forest Ecosystem Compensation Program (FECP) is a value compensation measure to improve the forest’s ecological benefits, protect and improve the ecological environment, and maintain the territorial ecological security in China.

Those policies aimed at reducing rural poverty through poverty alleviation resettlement, ecological compensation, and job employment in a series of ecological construction programs.

As farmer households’ behavior has an effect on their willingness to engage in farming and environmental governance [11], policies also have a similar effect on these behavioral responses, hence, the study of the impacts of these policies on the livelihood of farmer households has gained increasing attention from scholars. In addition, these behavioral responses also have a greater impact on selecting livelihoods than economic benefits, especially for farmers at the lower-income level [12]. Livelihood sustainability of farmer households has generally improved since the implementation of ecological compensation policies, as reflected by the increases in the sustainable livelihood capacity index, the adoption of modern agricultural practices, and employment in non-agricultural industries [13,14]. However, negative impacts on livelihoods have also been observed, particularly in the implementation of the Grain for Green Program (GGP), in which decreases in the incomes of farming households were recorded [15]. Meanwhile, improvements in the livelihood capital, patterns, and resilience were observed during the implementation of the Poverty Alleviation Relocation Program (PARP), hence, this has been considered to have a significant poverty reduction effect in its targeted localities [16,17].

The karst areas are one of the most ecologically fragile areas in China, with high numbers of the poor population in which approximately 9.23 million people belong to the absolutely impoverished population [18] in Guizhou Province. It has been the focus of national environmental management to restore severely damaged ecosystems and promote its karst landscapes in the future [19]. As its ecological problems are as serious as its poverty problems, regional ecological policies should play a significant role in guiding the livelihood decisions of local farmer communities [20]. A large proportion of its rural poor population lives in deep and rocky mountainous areas and in rocky desertification areas such as Guizhou. In this region, poverty alleviation is a top priority as per capita income has remained far below national socio-economic standards [21]. Poverty is linked to environmental degradation and vice versa, hence, measures to restore ecosystems combined with poverty reduction strategies that improve the livelihood of farmer households are critical for successful environmental sustainability and livelihood transformation in these eco-fragile areas.

Previous studies on the impacts of policies such as the GGP [15], Forest Eco-Compensation Program [22], Targeted Poverty Alleviation Policy [23] , and PARP [17,24,25] on the livelihoods of farming communities have focused primarily on their analysis at an individual and separate level, while there have been few that focused on multiple overlapping multiple policies in comparative studies. Meanwhile, livelihood studies focusing on livelihood changes and transitions in response to ecological policies, particularly the interactive feedback mechanism between the ecological policy system and farmer households’ livelihood change, have not been previously conducted, hence, this study aims to provide a foundation for future livelihood studies and contribute to the improvement and development of ecological policies for poverty alleviation in similar areas.

2. Research Data and Methods

2.1. Study Area

Anshun City has one of the largest concentrations of the typical karst land formation in the world. It is located in the midwestern portion of the Guizhou Province and is situated along the eastern slope of the Guizhou Plateau and in between the Yunnan Guizhou Plateau and Guangxi Hilly Basin. Among the cities in Guizhou, Anshun has one of the most severe soil erosion problems due to its highly mountainous features with steep slopes and shallow soils, accompanied by heavy and intense precipitation events. Hence, it has been identified as a critical erosion-prone area. To minimize erosion susceptibility, the local government of the Guizhou Province established three water and soil conservation zones. Based on the soil erosion classification of the area, the north central hills are classified as a light erosion zone, while the south plateau mountains are moderate erosion zones, and the southeast mountain valley basin is a moderate erosion zone. The local geology of the city is predominantly composed of 71.5% dissolved carbonate rocks, and such conditions have contributed to the rocky desertification phenomenon occurring in the area. This phenomenon is known as the “earth’s cancer” due to the poor natural conditions of the area with fragmented arable land and shallow soils; in addition, traditional agricultural practices (e.g., iron farm tools and cattle farming, especially slope-cultivated land) have also been widespread, causing serious soil erosion and rock desertification, which, in turn, result in long-term low and unstable crop yield and an over-capacity land population. Ultimately, these issues hinder the sustainable development of the economy and the well-being of its local communities. With this, Anshun City was selected as the study area for this study in an attempt to explain the relationship between eco-poverty alleviation policies and farmer household livelihoods (Figure 1).

Figure 1. The study area.

2.2. Data Collection

Primary data collection was conducted in two phases: Primary and supplementary surveys for some of the regions. The former was conducted from 20–30 September 2020, while the latter occurred in March 2021. To determine the ecological policy system and farmer household livelihoods, a survey using questionnaires was conducted in selected areas in Anshun City, in which the regional economic development status, eco-environmental conditions, and rural location were among the factors considered during the selection. Eight villages in five districts and counties were selected, in which specific areas included relocation sites for poverty alleviation, ethnic villages around A-level scenic spots for rural tourism development, demonstration sites of single-carbon-sink projects, returning farmland to forest areas, middle-income villages, low-income villages, and impoverished villages. As various areas of different characteristics were selected, sample selection was highly representative (Table 1). A total of 380 questionnaires were distributed of which 362 were considered usable while 18 were incomplete due to unanswered questions in the questionnaire survey. In addition, 16 on-site interviews with village cadres were also performed.

Table 1. Distribution of survey samples.

District/County	Township/Town	Village	Total Number of Households	Total Population	Number of Samples	Poverty Registration Households
Pingba	Qibo	Guankou	605	2292	56	29
Puding	Chuandong	Yutushan	1338	6799	86	27
Guanling	Guansuo	Xinchang	704	3062	45	6
	Shaying	Luzhi	476	2076	37	16
Zhenning	Huangguoshu	Shitouzhai	749	3321	46	7
	Geli	Gebang	314	1614	43	29
Ziyun	Bandang	Baiyang	305	1485	22	11
	Getuhe	Luoke	347	1697	27	4
Total			4838	22,346	362	129

Note: Poverty registration households refer to the poor families who have completed the approval process in the targeted poverty alleviation policies in China, established poverty files, incorporated them into the dynamic management of the national poverty alleviation and development information system, and obtained the ‘poverty alleviation manual’.

2.3. Methodology

2.3.1. The Models

Multivariate Logistic Model

In this study, a multinomial logistic regression model was used to analyze the impacts of the ecological policies on the farmer household livelihoods [26], in which the tri-categorical discrete variables (i.e., village-based livelihood pattern, in-province part-time, and out-of-province working patterns) were assigned as the dependent variables. Meanwhile, the “village-based livelihood pattern” was the control variable. The control variable was defined as Y = 1, whereas the “in-province part-time working pattern” was defined as Y = 2 and the out-of-province working pattern as Y = 3. The multivariate logistic model is formulated as:

$$I\mathrm{n}\left[A{(\mathrm{x}}_2)/A{(\mathrm{x}}_1)\right]=a_1+{\displaystyle \sum k=1{\beta }_{1k}{J}_k+\epsilon }$$

(1)

$$I\mathrm{n}\left[A(x_3/A(x_1)\right]=a_2+{\displaystyle \sum K=1{\beta }_{2k}{J}_k+\epsilon }$$

(2)

where A represents the probability of farmer livelihood pattern selection; X₁ is the choice of “village-based livelihood pattern”; X₂ is the choice of “in-province part-time employment pattern”; X₃ is the choice of “out-of-province working pattern”; ε is the random error; α_n is a constant term; J_K is the explanatory variable; and β_nk is the regression coefficient of the k_th influencing factor.

The Entropy Weight and TOPSIS Model

Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) is a method that has been widely used in the evaluation of policy implementation performances [27,28,29], in which it is used to rank and judge the merits of evaluation targets by setting the plus–minus ideal solutions of each evaluation index and obtaining the closeness degree [23]. In this study, the responsiveness of the farmers to the eco-poverty alleviation policy was considered indicative of its implementation performance. In addition, this method, in combination with the entropy coefficient method, can also determine any potential improvements in the policy.

(1)

Weight and entropy weight decision matrix

The weight is determined according to the information entropy of each index, in which the weight of the vector is M_ij = w_j = (w₁, w₂, …, w_n)^T.

$$R_{\mathrm{j}}=-k({\displaystyle \sum _{i=1}^m{p}_{ij}1n{p}_{ij}})$$

(3)

$$b_j=1-R_j$$

(4)

$$x_j=b_j/(n-{\displaystyle \sum _{j=1}^n{R}_j})$$

(5)

$$H_i=\frac{C_i^{-}}{C_i^{+}+C_i^{-}}(i=1,2\cdots ,m)$$

(6)

In Equations (3) to (6), b_j is the degree of information redundancy; R_j is the entropy of the jth index; k is the undetermined constant, k = 1/ln(m); the characteristic specific gravity of the indicator $p_{ij}=m_{ij}/{\displaystyle \sum m_i=1m_{ij}}$, when $p_{ij}=0$, set $p_{ij}1n{p}_{ij}=0$.

According to the normalization matrix and entropy weight coefficient, the weighted normalization decision matrix about the weighted normalization value $S_{ij}$ was obtained as follows:

$$S_{ij}=\begin{array}{cccc}{S}_{11}& S_{12}& \dots & S_{1j}\\ S_{21}& S_{22}& & S_{2j}\\ ⋮& & \ddots & ⋮\\ S_{i1}& S_{i2}& \cdots & S_{ij}\end{array}=w_j\cdot M_{ij}$$

(7)

(2)

Plus–minus ideal solution and closeness

The plus-ideal solution X⁺ and minus-ideal solution X⁻ are calculated using $S_{ij}$:

$$X^{+}=\left\{\max S_{ij}|j=1,2\cdots ,n\right\}=\left\{S_1^{+},S_2^{+},\cdots S_n^{+}\right\}$$

(8)

$$X^{-}=\left\{\min S_{ij}|j=1,2,\cdots ,n\right\}=\left\{S_1^{-},S_2^{-},\cdots S_n^{-}\right\}$$

(9)

The distances C⁺ of the plus-ideal and C⁻ of minus-ideal solutions from the farmer household rating vector of different livelihood patterns are computed as follows:

$$C^{+}=\sqrt{{\displaystyle \sum _{i=1}^n{\left(S_{ij}-S_j^{+}\right)}^2}}$$

(10)

$$C^{-}=\sqrt{{\displaystyle \sum _{i=1}^n{\left(S_{ij}-S_j^{-}\right)}^2}}$$

(11)

In Equations (10) and (11),$i=1,2,\cdots ,m$ C⁺ is the closeness between the evaluation vector of farmers with different livelihood patterns and the plus–ideal solution. The lower the value of C⁺, the higher the effect of farmers on ecological policies. Meanwhile, C⁻ is the closeness between the evaluation vector of farmers with different livelihood patterns and the minus-ideal solution.

(3)

The closeness of the plus-ideal solution was calculated as follows:

$$H_i=\frac{H_i^{-}}{H_i^{+}+H_i^{-}}(i=1,2\cdots ,)$$

(12)

In Equation (12), 0 ≤ H_i ≤ 1. Higher values of C_i indicate that the positive ideal solution is closer, which means that the response of farmer households to the eco-poverty alleviation policy is higher. Meanwhile, H_i = 0 indicates a low farmer response to the eco-poverty alleviation policy. Consequently, this means that the policy had not been recognized and adopted in the implementation process. H_i = 1 indicates that the implementation of the eco-poverty alleviation policy meets the expectations of the farmer household. The closeness was classified into four criteria levels to represent the responsiveness level of farmers following the implementation of the eco-poverty alleviation policy, which was also dependent on the actual conditions in the study area, in addition to previous research findings (Table 2).

Table 2. Levels of eco-poverty alleviation policy responsiveness.

Responsivity	Minimum Level	General Level	High Level	Higher Level
Closeness	0.0~0.29	0.3~0.59	0.6~0.79	0.8~1.0

2.3.2. Independent Variables

Impacts of the Eco-Poverty Alleviation Policy on the Farmer Livelihood Changes

Based on the relevant literature [30,31] and actual field investigations, this study identified three independent variables influencing farmer livelihood changes: Farm household livelihood capital, the perception of farmers, and the implementation of eco-poverty alleviation policies (Table 3). Meanwhile, the natural capital (household cultivated land area X₁, converted land area X₂), material capital (house area X₃, estimated value of household fixed assets X₄, number of trees participating in single-carbon-sink project X₅) social capital (community organization X₆, harmonious relationship with surrounding people X₇), financial capital (household average annual income X₈, type of income source X₉, bank loan X₁₀), and human capital (number of labor force X₁₁, family health condition X₁₂, poverty registration household X₁₃) were selected as the indicators of farmer household livelihood decisions. Farmer policy perception, which refers to formal and informal systems such as village regulations and traditional customs (X₁₄), and national ecological systems (X₁₅) were also considered, including the implementation of the eco-poverty alleviation policies, namely, GGP (X₁₆), PARP (X₁₇), EFAP (X₁₈), SCSP (X₁₉), FECP (X₂₀), etc.

Table 3. Variable descriptions.

Variables		Symbol	Variable Descriptions
Dependent variable	Farm household livelihood change or not	Y	Village livelihoods pattern = 1; In-province part-time employment pattern = 2; Out-of-province working pattern = 3
Natural capital	Household cultivated land area	X1	Less than 1 mu = 1; 1.1 mu~3 mu = 2; 3.1 mu~5 mu = 3; 5.1 mu~7 mu = 4; Over 8 mu = 5
	Converted land area	X2	Less than 1 mu = 1; 1.1 Mu~3 mu = 2; 3.1 mu~5 mu = 3; 5.1 mu~7 mu = 4; Over 8 mu = 5
Material capital	House area	X3	Less than 100 km2 = 1; 101 km2~300 km2 = 2; Above 301 km2 = 3
	Estimated value of Household fixed assets	X4	Less than RMB 500,000 = 1; RMB 510,000~1 million = 2; more than RMB 1.01 million = 3
	Number of trees participating in SCSP	X5	Less than 100 plants = 1; 101~300 plants = 2; 301~400 plants = 3
Social capital	Community organization	X6	Participate in the village committee, rural tourism and other relevant organizations or not: Yes = 1; No = 0
	Rapport with people around	X7	Whether the relationship with people around the village and colleagues is harmonious: Yes = 1, no = 0
Financial capital	Average annual household income	X8	Below 10,000 = 1; 11,000~30,000 = 2; 31,000~50,000 = 3; 51,000~70,000 = 4; over 80,000 = 5
	Type of income source	X9	Agricultural income = 1; Agricultural income + non-agricultural income = 2; Non-agricultural income = 3
	Bank loan	X10	Whether to obtain loans from banks and other financial institutions; Yes = 1; No = 0
Human capital	Number of labor force	X11	continuous variable
	Family health status	X12	Someone in the family is seriously ill or not: Yes = 1; No = 0
	Poverty registration household	X13	Belong to poverty registration household or not: Yes = 1; No = 0
Farm household perceptions	Impact of informal eco-poverty alleviation systems	X14	Village regulations and traditional customs have an impact on your choice of livelihood mode or not: Yes = 1; No = 0
	Impact of formal eco-poverty alleviation systems	X15	Whether the national ecological system has an impact on your choice of livelihood mode: Yes = 1; No = 0
Implementation of eco-poverty alleviation policies	GGP	X16	Whether it is a household GGP: Yes = 1; No = 0
	PARP	X17	Whether it is a household of PARP: Yes = 1; No = 0
	EFRP	X18	EFRP or not: Yes = 1; No = 0
	SCSP	X19	Whether it is a participant of SCSP: Yes = 1; No = 0
	FECP	X20	Whether there is FECP: Yes = 1; No = 0

Note: 1 mu = 0.667 ha.

Farmer Responses to Household Livelihood Changes

The Social Cognitive Theory (SCT) is a learning theory that has been widely used in studying behavioral intentions and choices at the individual level [32]. It posits three elements, namely, the sense of self-efficacy, value expectation perception, and external environment perception [33]. SCT is widely used in some studies of individual behavior intention and behavior choice, and the response to the implementation process of ecological policy mostly depends on the subjective cognition of farmer households. Therefore, it was used to understand the responses of farmer households to the implementation process of the studied eco-poverty alleviation policies, which focuses primarily on the subjective perception of farmer households. The policy response evaluation index system is primarily constructed based on social cognitive theory.

The evaluation index system used to determine the degree of policy responses of farmers was based on the premise of the elements of SCT (Table 4). For instance, as farmers generally recognize and evaluate their own abilities and risk exposures, it reflects their sense of self-efficacy. In addition, their own perceptions of eco-poverty alleviation policies and their experiences in participating in these also contribute to their sense of self-efficacy. With this, two indicators reflecting farmer perception and risk cognition were selected. Meanwhile, when individuals make a behavioral selection based on the expectation of favorable results, this reflects their value expectation, as farmers tend to focus more on maximizing results for the interests of themselves and their families prior to making any decision. When individuals consider specific situations in their decisions, their behavior is reflected by their external environment [13].

Table 4. Index system of farmer household responses to eco-poverty alleviation policies.

Target Layer	System Layer	Factor Layer	Index Layer	Value
Eco-poverty alleviation policy responsiveness	Self-efficacy (S)	Content perceptions (S1)	Understanding degree of ecological policies (S1-1)	1~5
			Importance degree of eco-poverty alleviation policies to poverty alleviation (S1-2)	1~5
			Importance degree of eco-poverty alleviation policy to rural revitalization (S1-3)	1~5
		Risk cognition (S2)	Reduction of land area after returning farmland to forests (S2-1)	1~5
			Transaction volume of single carbon sink project (S2-2)	1~5
			Ecological forest ranger or not (S2-3)	1~5
	Value expectation (V)	Family expectation (V1)	Increase in average annual household income (V1-1)	1~5
			Significant improvement in family life (V1-2)	1~5
			Substantial increase in agricultural income (V1-3)	1~5
			Rapid transformation of livelihood pattern (V1-4)	1~5
	External environment (E)	Natural capital (E1)	Quantity and quality of cultivated land (E1-1)	1~5
		Material capital (E2)	Household fixed assets (E2-1)	1~5
		Financial capital (E3)	Household non-agricultural income (E3-1)	1~5
		Social capital (E4)	Neighborhood support (E4-1)	1~5
		Human capital (E5)	Household labor force (E5-1)	1~5

3. Results

3.1. Implications of Livelihood Capital on Changes in Farmer Livelihoods

The different variables used in this study were found to have different influences on the changes in farmer household livelihoods (Table 5). For instance, natural capital such as cultivated land areas (X₁) were found to be insignificant in model 1, whereas the Estimated value of Household fixed assets (X₄) (B = 5.569, p < 0.05) and financial capitals such as average annual household incomes (X₈) (B = 1.654, p < 0.05) were significant in transforming farmer households’ livelihood. This means that higher household incomes equate to higher capacities to invest in non-agricultural industries, hence, the higher the likelihood of farmers selecting in-province part-time working patterns. As income sources (X₉) were also significant in model 1, this indicates that a typical single income source was also found to have transitioned into multiple and diversified income sources, in which their livelihood patterns have also been converted from farming to various part-time working patterns.

Table 5. Impacts on changes in farmer household livelihoods.

Variables	In-Province Part-Time Employment Pattern (Model 1)			Out-Of-Province Working Pattern (Model 2)
	B	S.E	Wald	B	S.E	Wald
X1	1.652	11.253	1.251	5.387 *	2.255	5.6871
X2	2.565	0.356	1.055	1.256	0.368	0.492
X3	0.657	0.848	0.566	0.627	0.299	2.471
X4	6.254	12.652	1.055	5.569 **	1.254	15.265
X5	0.258	0.487	3.011	3.324	1.214	13.241
X6	0.542 ***	0.254	0.086	0.478	0.457	0.036
X7	0.256	0.248	0.745	0.965	0.874	0.256
X8	1.654 **	0.815	0.273	1.452 *	0.478	0.687
X9	1.253 **	0.268	1.054	0.514 *	0.468	0.257
X10	0.924	0.039	14.231	0.474	0.657	16.234
X11	0.211 ***	0.454	0.056	0.312 ***	0.654	0.054
X12	1.354 **	0.421	2.214	1.142 **	3.245	0.246
X13	0.235 *	0.542	1.236	0.218 *	1.685	0.425
X14	0.367 **	1.054	0.957	1.238	0.874	0.654
X15	1.263 **	0.247	0.857	1.542 **	1.245	0.544
X16	0.563	0.425	0.685	0.356	0.456	0.235
X17	1.105 ***	0.254	2.475	1.235 ***	0.563	0.258
X18	0.498 **	0.156	0.985	0.397	0.236	0.572
X19	0.329	1.051	0.451	0.364	0.456	0.254
X20	0.265	0.564	0.875	0.298	0.384	0.564
−2log likelihood = 254.124             LR chi2 (33) = 241.537     Prob > chi2 = 0. 0000

Note: ***, **, * indicate significance levels under 0.01, 0.05, and 0.1, respectively.

Number of labor force (X₁₁) and family health status (X₁₂) were significant in both Model 1 and Model 2. Number of labor force is one of the main factors of family income. Agricultural income makes it difficult to feed family members with a large labor force, so such families must find another way to promote the probability of their livelihood change.

Meanwhile, family health status was found to have an impact on the livelihood transformation, in which “illness” or “disability” were the primary reasons that impoverished households remain poor. With this, they are less likely to change their livelihoods.

As the diversity of farmer household income sources is linked to the diversity of their livelihood, farmer household perceptions on the impact of the ecological informal system on village regulation and traditional customs (X₁₄) (B = 0.367; p < 0.05) were found to be significant in model 1. This indicates that the livelihood conversion to in-province part-time working patterns is affected by the informal ecological system. This was in contrast to the actual survey results where the age of the farmer household samples had significant effects instead. In addition, the perception of farmers of the impact of the national ecological system (X₁₅) was significant in both models (B = 1.263, p < 0.05; B = 1.542, p < 0.05), indicating that farmer perception has a significant influence on ecological policies, which, likewise, impact farmer livelihood transitions.

PAG (X₁₇) had a significant impact on the livelihood change of farmers, as shown by the significance found in models 1 (B = 1.105, p < 0.01) and 2 (B = 1.235, p < 0.01). GGP (X₁₆) and EFRR (X₁₈) were significant in model 1 (B = 0.498, p < 0.05); however, both were not significant in model 2, because, owing to the fixed working hours and locations of the EFRP and provisions of national financial subsidies, they choose village-based livelihood patterns more often instead of in-province part-time employment patterns and out-of-province working patterns. SCSP (X₁₉) and EFBC (X₂₀) did not have significant impacts, which may be attributed to the number of participating farmers. The alleviation policies with the highest impact on changing farmer livelihood were in the following order: PARP (X₁₇) > EFRP (X₁₈) > GGP (X₁₆) > FECP (X₂₀) > SCSP (X₁₉).

3.2. Policy Responsiveness of Farmers with Different Livelihood Patterns

3.2.1. Policy Response Statistics of Farmers

Farmer households with different livelihood patterns responded differently to the studied policies (Table 6). In terms of the mean size of indicators, the village-based livelihood pattern obtained the highest value of 3.546, followed by the part-time in-province working pattern with 3.268 and the out-of-province working pattern with 2.811. Based on the value expectation indicators (X₇~X₁₀), farmer households with village-based livelihood patterns had the highest policy expectations with 3.337, followed by the part-time in-province part-time working pattern (3.156) and the out-of-province working pattern (2.965). It was observed that farmer households with village-based livelihood patterns with a small-scale livelihood range have the highest expectations of these policies, in which they expect their livelihoods can be improved through its implementation. Meanwhile, farmer households with out-of-province working patterns and diversified livelihoods had low expectations. Their dependence and awareness of the policies were low as their income sources were not generated from agricultural livelihoods.

Table 6. Descriptive statistics of the policy responses of farmers with different livelihood patterns.

Index	Village-Based Livelihood Pattern		In-Province Part-Time Employment Pattern		Out-Of-Province Working Pattern
	Mean Value	Standard Deviation	Mean Value	Standard Deviation	Mean Value	Standard Deviation
X1	3.784	0.801	4.325	1.023	2.654	1.135
X2	4.211	0.962	2.458	1.123	3.568	1.087
X3	3.654	0.957	3.575	0.965	2.354	0.958
X4	2.451	1.121	2.245	1.356	1.562	0.542
X5	3.524	1.595	4.356	0.989	2.554	1.159
X6	4.059	1.059	3.562	0.896	1.567	1.054
X7	3.785	0.806	2.456	1.032	3.695	0.884
X8	3.565	0.921	3.698	1.632	2.221	1.036
X9	2.541	1.154	4.356	1.235	2.354	1.436
X10	3.456	1.425	2.112	1.156	3.591	1.071
X11	3.964	0.851	3.265	0.889	2.442	1.032
X12	3.875	0.843	2.654	1.534	3.657	1.514
X13	4.121	0.912	4.321	0.965	4.123	0.901
X14	2.624	1.456	3.256	0.996	2.102	1.001
X15	3.582	1.611	2.382	1.356	3.724	0.862
Mean value	3.546	1.098	3.268	1.165	2.811	1.045

3.2.2. Policy Responsiveness of Farmer Households

Farmer households with different livelihood patterns had varying responses to the different eco-poverty alleviation policies (Table 7). The general response level of farmers to different ecological policies had a degree of 0.578. Meanwhile, their responses to GGP and PARP were both high, with an average response of 0.665 and 0.776, respectively. Among farmers, those working out-of-province had the highest response to GGP with 0.685, while those with village-based livelihood patterns had the lowest. The farmer household responses to SCSP were below average with an average response of 0.289. Here, farmers with village-based livelihood patterns responded slightly higher than those with part-time in-province and out-of-province working patterns, with 0.296. The general response level of farmer households to the EFRP was 0.58. Farmers with village-based livelihood patterns had the highest response to this policy with 0.603. Based on this, the livelihood activities and patterns of farmer households influence their policy responsiveness, in which farmer households with a village-based livelihood pattern had higher responsiveness than those with a part-time in-province and part-time working pattern, due to the existence of agricultural income in the village and non-agricultural income in out-of-province areas. This indicates that the income sources of those with an out-of-province working pattern are primarily generated from non-agricultural activities, while income sources of farmer households with village-based livelihood pattern income primarily come from agricultural production activities.

Table 7. Responsiveness and closeness of farmers with different livelihood patterns to different eco-poverty alleviation policies.

Patterns	C+	C−	Closeness	Responsiveness	Response Level
GGR
Village-based livelihood pattern	0.214	0.154	0.314	0.652	High
In-province part-time employment pattern	0.231	0.163	0.332	0.657	High
Out-of-province working pattern	0.221	0.145	0.346	0.685	High
Mean value	0.222	0.154	0.331	0.665	High
PARP
Village-based livelihood pattern	0.374	0.185	0.546	0.865	Higher
In-province part-time employment pattern	0.354	0.175	0.468	0.742	High
Out-of-province working pattern	0.349	0.173	0.457	0.721	Higher
Mean value	0.359	0.178	0.490	0.776	Higher
SCSP
Village-based livelihood pattern	0.221	0.102	0.186	0.296	Low
In-province part-time employment pattern	0.204	0.112	0.187	0.285	Low
Out-of-province working pattern	0.206	0.134	0.175	0.286	Low
Mean value	0.210	0.116	0.183	0.289	Low
EFRP
Village-based livelihood pattern	0.286	0.152	0.479	0.603	High
In-province part-time employment pattern	0.265	0.135	0.474	0.586	Medium
Out-of-province working pattern	0.271	0.142	0.421	0.551	Medium
Mean value	0.274	0.143	0.458	0.580	Medium

3.2.3. Responsiveness to the Eco-Policies of Farmer Households in Different Regions

Various responses to each eco-poverty alleviation policy were observed in farmer households from different regions, in which farmers residing in hilly erosion zones obtained the highest response with an average of 0.573, while farmers from the plateau and mountainous erosion areas obtained the lowest with 0.543 (Table 8). Among the studied policies, the farmer households responded highest to GGP, with 0.637, while SCSP obtained the lowest response with 0.491. Farmer responses to the studied policies were in the following order: GGP > PARP > EFRP > SCSP. Specifically, farmers from the Pingba District of the hilly erosion zone had the highest response to SCSP with 0.634.

Table 8. Regional farmer household responsiveness to eco-poverty alleviation policies in different regions.

Area	Districts /Counties	Sample Village	C+	C−	Closeness	Responsiveness	Response Level	C+	C−	Closeness	Responsiveness	Response Level
			GGP					PARP
Hilly erosion area	Pingba	Guankou	0.396	0.102	0.512	0.634	High	0.285	0.123	0.426	0.521	Medium
	Puding	Yutushang site	0.311	0.106	0.518	0.643	High	0.275	0.105	0.401	0.514	Medium
Plateau mountain erosion area		Linggang	0.376	0.113	0.527	0.652	High	0.291	0.129	0.428	0.524	Medium
	Guanling	Luzhi	0.321	0.098	0.501	0.604	High	/	/	/	/	/
		Xinchang	0.365	0.106	0.515	0.637	High	/	/	/	/	/
	Zhenning	Shitou	0.321	0.096	0.498	0.601	High	/	/	/	/	/
Mountain valley basin erosion zone		Gebang	0.385	0.124	0.534	0.675	High	0.301	0.135	0.456	0.576	Medium
	Ziyun	Baiyang	0.365	0.109	0.521	0.645	High	0.295	0.145	0.435	0.535	Medium
		Luoke	0.379	0.108	0.515	0.638	High	0.246	0.123	0.446	0.567	Medium
Mean value			0.358	0.107	0.516	0.637	High	0.282	0.127	0.432	0.540	Medium
			SCSP					EFRP
Hilly erosion area	Pingba	Guankou	0.365	0.126	0.523	0.653	High	0.245	0.135	0.439	0.542	Medium
	Puding	Yutushang	/	/	/	/	/	0.234	0.142	0.395	0.501	Medium
Plateau mountain erosion area		Linggang	0.361	0.163	0.432	0.562	Medium	0.262	0.125	0.406	0.512	Medium
	Guanling	Luzhi	/	/	/	/	/	0.271	0.144	0.442	0.562	Medium
		Xinchang	0.298	0.124	0.165	0.278	Low	0.215	0.136	0.398	0.501	Medium
	Zhenning	Shitou	/	/	/	/	/	/	/	/	/	/
		Gebang	0.316	0.122	0.354	0.425	Medium	0.212	0.152	0.412	0.512	Medium
Plateau mountain erosion area	Ziyun	Baiyang Village	0.391	0.196	0.496	0.564	Medium	0.	0.188	0.435	0.521	Medium
		Luoke	0.366	0.165	0.365	0.465	Low	/	/	/	/	/
Mean value			0.350	0.149	0.389	0.491	Low	0.240	0.146	0.418	0.522	Medium

3.2.4. Obstacle Factor Diagnosis Farmer Responsiveness to Various Policies

As shown in Table 9, the obstacle factors in the responsiveness of farmers to the eco-poverty alleviation policies were primarily related to the different livelihood types, and among the regions, three different regions exhibited similar results. Of the three response indicators, the external environment system was observed to have the highest impact. Simultaneously, it also had the highest obstacle degree, in which the three village-based livelihood patterns obtained the highest obstacle degree with 66.57, 66.57, 66.51, and 50.80, respectively.

Table 9. Obstacle degree of farmers’ policy response to different livelihood patterns in different regions.

Region	Village-Based Livelihood Pattern								In-Province Part-Time Employment Pattern								Out-of-Province Working Pattern
	S1	S2	V1	E1	E2	E3	E4	E5	S1	S2	V1	E1	E2	E3	E4	E5	S1	S2	V1	E1	E2	E3	E4	E5
Hilly erosion area	2.87	25.23	5.62	2.74	15.1	35.6	0.33	12.8	2.35	23.26	5.24	2.91	14.31	32.46	0.32	16.51	0.03	23.26	5.14	2.71	13.31	30.23	0.23	4.32
Plateau mountain erosion area	2.56	24.64	5.36	2.23	13.20	37.23	0.56	13.86	2.25	26.96	4.63	2.35	15.62	38.61	0.63	15.23	0.12	26.25	4.23	3.25	14.23	35.56	0.13	4.02
Mountain valley basin erosion area	2.63	26.06	5.95	2.01	12.23	37.68	0.78	14.56	2.56	27.56	4.35	2.46	16.45	37.53	0.56	11.21	0.18	27.22	4.05	3.24	15.01	39.05	0.04	4.05

Of the eight variables, the highest obstacle degree was observed in the financial capital (E₃). The village-based livelihood pattern obtained a degree of 7.68, which was highest in the mountain valley basin erosion area. The in-province part-time working pattern obtained a maximum obstacle degree of 38.61, particularly in the highland mountain erosion area. Although farmer households in the village-based livelihood pattern have a relatively good understanding of relevant ecological policies, their education level was observed as a hindrance to their awareness and responsiveness to the policies, as observed in farmers under SPCSP. Based on field observations, farmers perceive that monetary income cannot be achieved without environmental destruction such as tree harvesting. In addition, farmers were also relatively unaware of the concept of carbon emission and carbon neutralization, more so in the SCSP policy.

4. Discussions and Conclusions

In this study, the impact of four eco-poverty alleviation policies on the changes in farmer household livelihoods was analyzed based on a survey conducted from 2019 to 2021 in different villages in Anshun City to determine the response mechanism of the impoverished population in the karst area of China to these policies. The impact of policies on the livelihoods of farmer households was determined using a multinomial logistic regression model, in which the entropy weight and TOPSIS method were used in combination with the entropy coefficient method.

It was found that eco-poverty alleviation policies have an impact on how farmers select their livelihoods and working patterns. Specifically, the implementation of eco-poverty alleviation policies resulted in changes in their livelihood capital. The availability of land for cultivation (natural capital) was also determined to have a minimal effect on their livelihood transition, and due to the rising cost of land cultivation and maintenance accompanied by the low return of investment due to low prices of agricultural products, farmers have opted to cease their agricultural livelihood in place of employment in urban areas. With less natural capital, farmers are forced to accumulate more assets, especially skills and specializations, so they can opt for part-time employment outside the province.

The results show that the impact of GGP on changing the livelihoods of farmer households was not significant in the study area. This may be attributed to the position of pertinent ecological systems in the area, resulting in the further decline of the perception and awareness of the farmers regarding its impacts. PARP was found to have positive effects on the livelihoods of farmers with their relocation from environmentally degraded areas to urban areas for resettlement. Such migration enabled their increased participation in non-agricultural activities, thereby improving their livelihoods. Their perception of eco-poverty alleviation policies was also apparent as they had significant impacts on transforming livelihoods.

PARP and EFRP policies create and provide employment opportunities for poverty registration households. There are also options for the farmer household livelihood model and a range of activities. EFRR increased poverty registration households’ incomes at their doorstep, which not only can effectively solve the social problems such as empty-nesters and left-behind children brought by migrant workers, but also provide employment opportunities for farmer households with the in-province part-time employment pattern. By 2020, the number of EFRPs in Guizhou reached 172,500 people. According to the standard that each ecological forest ranger earns RMB 10,000 a year to lift three poor people out of poverty, EFRP has lifted 517,500 people out of poverty in Guizhou (http://www.xinhuanet.com/ accessed on 8 January 2023). It built a better ecosystem in poverty-stricken areas in Guizhou Province and perfected the eco-compensation mechanism. On top of that, the province stays committed to improving the forestry sector so that impoverished people will benefit more from constructing an ecological system.

The recent improvements in the transportation patterns in Guizhou also resulted in an increase in livelihood opportunities for farmer households, thereby alleviating poverty and revitalizing rural living. In particular, it was observed that farmers have increased involvement in primary industries to secondary and tertiary industries, which enabled the diversification of their livelihoods from purely agricultural to diversified livelihood patterns.

As the Guankou Village in Pingba District was the pilot batch of the poverty alleviation project launched in Guizhou Province by SCSP, the response levels in plateau mountainous and mountain valley basin erosion zones were relatively low. Here, a new poverty alleviation model adopted the “internet + ecological construction + targeted poverty alleviation” strategy to contribute to poverty alleviation and ecological restoration, and the awareness of local farmers was relatively high, including their willingness to participate. However, awareness, willingness to participate, and responsiveness were all low in other regions.

The result of this study provides strong evidence that the obstacle factor diagnosis of the responsiveness of farmer households to eco-poverty alleviation policies varies significantly depending on the region. As the geographical location determines the economic development of a region, it has a direct influence on the recognition and behavior of farmer households. In this study, the degree of policy obstacles in the hilly erosion area, which was in close proximity to the municipal government and accompanied by good traffic conditions, was significantly lower than that of the other two regions. Notably, there were common characteristics in the livelihood patterns among these regions, in which all are mountainous areas with poor land resources, agricultural production and agricultural product prices are low, the majority of households rely on employment outside the home, and non-agricultural livelihood is the main source of household income. With these, it is inferred that the environment of farmer households also affects their degree of response to ecological policies.

To optimize the benefits of eco-poverty alleviation policies and livelihood changes for farmer households, this study recommends the following: (1) The government should strengthen the promotion and publicity of SCSP, mobilize the participation of poor households, monitor ground conditions of the carbon sequestration pilot in poor villages, and continuously expand the coverage of the pilot to more impoverished villages and households; (2) the participation of farmer households should be enhanced by improving their awareness of these policies, including their implementation measures; (3) subsidy standards of farmer households of GGP should be increased, especially in fragile ecological areas, and the eco-poverty alleviation policies must reflect the conditions of the different regions to incorporate and streamline those differences. Lastly, governments should also provide adequate guidance to these farmers to actively transform their livelihoods for the improvement of their quality of living.