The Impact of Forestry Technological Innovation on the Welfare of Farm Households Managing Jujube Forests (Ziziphus jujuba Mill.) in the Lüliang Mountains of the Yellow River’s Middle Reaches

Abstract

Jujube (Ziziphus jujuba Mill.) makes up a traditional characteristic industry with ecological significance in the Lüliang Mountain of middle reaches of the Yellow River (LMMRYR). However, low economic efficiency has reduced local farm households’ willingness to continue jujube cultivation, threatening the sustainable maintenance and development of jujube forests and the ecological environment. In response, Lüliang City implemented a technological innovation program, that is, the Jujube Forest High Grafting and Optimization Program (JFHGOP), in 2018. Based on survey data from 302 local farm households, an empirical analysis using propensity score matching and ordinary least squares methods revealed that the program significantly enhanced the economic, ecological, and social benefits for participating farm households, improving their overall welfare. Robustness tests confirmed these findings, and a heterogeneity analysis showed varied impacts across different dimensions. The program improved welfare through government support and cooperatives’ assistance. To further promote green development and farm households’ welfare, recommendations include advancing forestry innovation technology, supporting small farm households with policy, capital, and technology, optimizing subsidy mechanisms, supporting new business entities, and promoting cooperation and benefit-sharing among stakeholders.

1. Introduction

Lüliang Mountain (LM), located in the middle reaches of the Yellow River, is characterized by its concentrated hills and valleys, with few flat plains and basins. It is one of the most severely eroded areas in the Yellow River Basin, with delicate and sensitive ecosystems. Ecological conservation in this region is crucial for the ecological security of the Yellow River’s middle and lower reaches. At the same time, LM faces challenges of low rural income. Thus, the dual issues of ecological fragility and economic underdevelopment present significant challenges for the LMMRYR.

Jujube (Ziziphus jujuba Mill.) forests are crucial both ecologically and economically for preventing soil erosion in the Yellow River Basin [1]. With nearly 3000 years of cultivation in the Lüliang Mountain (LM) region, jujube trees hold significant ecological and cultural value, while providing a vital income source for local farmers. The jujube industry, a hallmark of the LM region, is seen by the local government as a key tool for balancing ecological and economic development. Lin County, a major jujube cultivation hub, features approximately 54,667 hectares of jujube forests, representing 26% of the province’s total jujube acreage. Jujube trees are distributed across 17 townships and 454 villages, involving around 370,000 jujube-farming households [2]. In recent years, the effectiveness of jujube cultivation has declined due to aging varieties [3,4] and issues such as funding, technology constraints, and price fluctuations [5]. Consequently, farm households’ willingness to engage in cultivation has decreased, hindering local green development and farm household welfare. In response, a collaborative effort involving the Chinese Society of Forestry, China Academy of Forestry Sciences, Hebei Agricultural University, Beijing Forestry University, and Lin County governments established the JFHGOP in 2018. This program aimed to explore new development pathways for jujube forests and address challenges such as low farm household engagement and profitability. The program employed techniques (e.g., single-head single-ear, single-head multi-ear, and multi-head multi-ear grafting) and utilized methods (e.g., split grafting, belly grafting, and bark grafting) to update traditional varieties with newer ones such as the “Linhuang No. 1”. Additionally, the program encouraged understory management to enhance the benefits of jujube cultivation. The JFHGOP has significantly influenced local areas by advancing forestry technology and promoting green development. Given that jujube cultivation in Lin County is vital for increasing farm incomes and contributing to soil and water conservation, evaluating its role and impact is of significant socioeconomic importance.

Extensive research has been conducted on various aspects of the Chinese jujube industry, including industrial chain integration, development bottlenecks, and future opportunities and challenges. Previous studies have also explored pathways and strategies for the transformation and upgrading of the jujube industry [6,7,8]. In terms of jujube industry technology, there are increasing studies on cultivation techniques, variety breeding, pest and disease control, product preservation, and deep processing [9,10]. However, jujube forest variety breeding is still in its early stages [11], with production largely relying on traditional local varieties [12,13]. Research on high grafting and optimization techniques for jujube forests has primarily focused on technical aspects [14,15], with limited studies comprehensively analyzing the impacts of these techniques on the economic, ecological, and social welfare of farm households.

Research on the jujube industry in the LM region primarily focuses on industry development issues and corresponding strategies [3,16,17]. Some studies have addressed the economic and ecological benefits of jujube production [18], utilizing methods such as cost–benefit analysis, comparisons of different jujube varieties and cultivation land types [19], and recommendations for classified management [4]. This study adopted both a quasi-natural experimental approach and empirical regression analysis methods to investigate the comprehensive welfare effects of the JFHGOP in the LM region. It is significant for clarifying the relationship between technological innovation in jujube cultivation and farm household welfare in this area.

In summary, while some scholars have analyzed the development issues and prospects of the jujube industry from various perspectives, none have comprehensively addressed the economic, ecological, and social welfare of farm households. Additionally, technological innovation as a “new productive force” has received insufficient attention. This paper addresses these gaps by examining the JFHGOP launched in Lin County, LM region, in 2018. Using micro-survey data from 302 farm households, we applied propensity score matching and ordinary least squares regression analysis to study the comprehensive welfare impact of technological innovation in jujube forests. We also conducted heterogeneous analysis based on different household characteristics to explore how technological innovation affects farm household welfare.

The marginal contributions of this paper are as follows: (1) providing a thorough analysis of the ecological, economic, and social welfare impacts of jujube forest management through the JFHGOP. (2) Focusing on the LMMRYR, with particular attention to forestry green development in ecologically fragile and economically underdeveloped areas. (3) Analyzing how the JFHGOP impacts farm household welfare with support from governments and cooperatives.

2. Theoretical Analysis and Research Hypotheses

2.1. The Welfare Effects of the JFHGOP on Farm Households

The JFHGOP aims to enhance the welfare of local farm households, with a strong emphasis on technology extension as its core strategy. Generally, promoting technology in rural areas has been shown to boost income and improve the quality of life [20,21]. By introducing technology, knowledge, and skills to farmers, their human capital can be significantly enhanced [22]. Through learning and sharing effects, farmers and other agricultural entities can continuously improve their capacity for technology transfer and diffusion, leading to optimized production methods, improved technical skills, and expanded income sources. Initially, the LM area followed a path of scale expansion and extensive management [23], with relatively low levels of technology and management practices [24]. As shown in Figure 1, the JFHGOP, through the promotion and application of technology, aims to reduce production risks for farm households and improve their return on investment [25], thereby increasing economic benefits. Additionally, it is expected to boost farm household productivity, promote the sustainable management of jujube forests, and enhance their ecological and social benefits [26,27], leading to an improved quality of life for farm households. Therefore, this paper proposes the following hypotheses:

H1. 

The JFHGOP can increase farm households’ economic benefits.

H2. 

The JFHGOP is also expected to improve local ecological and social benefits, thereby promoting farm household welfare.

2.2. The Pathways of the Welfare Effects of the JFHGOP on Farm Households

Given that the JFHGOP can enhance the welfare of farm households in economic, ecological, and social aspects, how does it contribute to the improvement of farm household welfare? This paper elucidates this through two pathways: government support and cooperatives’ assistance (Figure 1).

Government support remains a crucial approach for revitalizing industries in impoverished and remote mountainous areas. Such support includes subsidies, loan assistance, relevant technical training, market information, and other forms of assistance [28]. Among these, subsidies are the commonly used policy tool by the government to protect and promote industry development, aiming at supporting and encouraging operations [29]. The promotion of the JFHGOP has a clear government-driven aspect. On the one hand, the government can reduce farm households’ production input costs and increase their income by providing charge-free technical training and subsidies for agricultural inputs, thus alleviating the financial pressure and technical difficulties faced by households in jujube forest management. On the other hand, government support sends a definite signal of support and development for the jujube industry, encouraging the production enthusiasm of farm households, cooperatives, and other operating entities, and enhancing the level of technical extension services. Meanwhile, jujube forests have ecological functions and positive externalities. Under government promotion, farm households’ individual choices are consistent with the practical needs of ecological benefits. Additionally, the increase in farm households’ production enthusiasm may allow them to increase agricultural labor time, obtaining non-cash benefits such as familial affection and a sense of belonging [30], thereby enhancing farmers’ social benefits. Accordingly, this paper proposes Hypothesis 3:

H3. 

Government subsidies enhance the welfare effects of the JFHGOP.

The impact of cooperative assistance on farm household welfare manifests in three main aspects: the integration of production factors, educational functions, and emotional support [31,32]. Firstly, cooperatives can achieve the effective allocation of production factors by promoting land scale management, intensifying capitals, and organizing labor resources [31]. This scale effect, along with the JFHGOP, can boost farm household incomes, increase property income through land circulation in jujube forests, and raise wages through intra-cooperative work. Secondly, cooperatives play an important role in education and training, helping households recognize the value of ecological conservation forests [33]. Finally, the emotional support provided by cooperatives fosters a sense of achievement and belonging, enhancing the peace of mind and spiritual well-being of farm households participating in the cooperatives [34]. Intra-cooperative work also meets the demand for local households, contributing to their sense of happiness in life.

H4. 

Cooperatives enhance the welfare effects of the JFHGOP.

3. Materials and Methods

3.1. Data Sources

The data for this study were collected through a survey conducted by the research team from 21 July 2022 to 1 August 2022, in Lin County, Shanxi Province. This survey, part of the JFHGOP, covered four towns in Lin County. We surveyed a total of 310 farm households, with 302 households providing usable data—157 participating and 145 non-participating.

To select participants, we first identified the towns and villages where the program was implemented. Data were gathered from participating households in these towns, specifically in Baizhai Village, Gaojiazui Village, Xuejiagetai Village, Zhaizeping Village, Xiwan Village, and Houtaizhen Village. Due to COVID-19 pandemic restrictions, we aimed to survey as many households as possible, ultimately collecting data from 157 participating households.

According to standard program evaluation methods, it is preferable for the sample characteristics of participating and non-participating households to be similar. Therefore, we also sampled non-participating households within the program villages and nearby villages, such as Haojiata Village and Luojiashan Village, resulting in data from 145 non-participating households. The empirical analysis was based on the data from these 302 farm households (

Table 1. Distribution of the sample of farm households.

Sample Town	Sample Village	Number of Farm Households		Percentage of Households (%)		Total (%)
		Participating Household	Non-Participating Household	Participating Household	Non-Participating Household
Shibaitou Township	Baizhai Village	13	0	8.28	0	4.30
	Gaojiaju Village	26	8	16.56	5.52	11.26
Linjiaping Town	Xuejiagetai Village	30	20	19.11	13.79	16.56
	Haojiata Village	0	21	0	14.48	6.95
Sanjiao Town	Luojiashan Village	0	35	0	24.14	11.59
Qikou Town	Zhaizeping Village	40	35	25.48	24.14	24.83
	Xiwancun Village	8	21	5.10	14.48	9.60
Qikou Town	Houtai Town Village	40	5	25.48	3.45	14.90
Total		157	145	100	100	100

).

3.2. Key Variables and Their Measurement

3.2.1. Key Independent Variable

The key independent variable chosen for this study was whether the farm households participated in the JFHGOP. As shown in

Table 2. Descriptive statistical analysis and test of differences in sample means.

Variable	Variable Definition and Unit	2017		2021		2017
		Control Group (n = 145)	Treatment Group (n = 157)	Control Group (n = 145)	Treatment Group (n = 157)	Mean Difference (t-test)
Involvement in programs	0 = not involved 1 = Participation	0	1	0	1
Jujube forest operation income	CNY	355.262	698.072	12.721	110.214	−342.810 **
Jujube forest understory operation income	CNY	178.386	327.809	122.616	1471.700	−149.423
Other agricultural and forestry operation income	CNY	1036.245	3786.892	904.728	3796.473	−2750.647 **
Jujube forest land transfer income	CNY	0	0	106.577	886.669
Jujube employment income	CNY	0	0	175.747	119.179
Biodiversity	1 = significant decrease 2 = minor decrease. 3 = no change 4 = minor increase 5 = significant increase			3.469	3.522
Land resource conservation	Same as above			3.221	3.350
Forest landscape	Same as above			3.221	3.350
Employment opportunities	Same as above			3.186	3.312
Life satisfaction	Same as above			3.834	3.911
Gender	0 = Female, 1 = Male	0.497	0.656	0.497	0.656	−0.159 **
Age	year	60.138	58.076	64.138	62.076	2.061 *
Education level	0 = Illiterate 1 = Primary school 2 = Junior high school 3 = High school 4 = Associate degree and above	1.297	1.522	1.297	1.522	−0.226**
Leader	0 = No, 1 = Yes	0.021	0.057	0.021	0.057	−0.037
Internet	0 = No, 1 = Yes	0.552	0.701	0.552	0.701	−0.149 ***
Household size	Individuals	1.993	2.236	1.993	2.236	−0.243 *
Labor force size	Count	0.921	1.232	0.917	1.242	−0.312 ***
Migrant labor force size	Count	0.434	0.478	0.490	0.465	−0.043
Arable land area	Hectares	0.490	0.576	0.490	0.577	−1.342
Rural land contract management certificate	0 = no, 1 = yes	0.910	0.975	0.910	0.975	−0.064 **
Jujube technical training	0 = no, 1 = yes	0.151	0.318	0.151	0.318	−0.167 ***
Government subsidy	Yuan	0	0	20.514	35.377

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively; Ziziphus jujuba Mill. is the scientific name for the jujube tree, which is commonly referenced in this study for income analysis related to its cultivation; the mean differences were calculated by subtracting the corresponding indicators of non-participant households from those of participant households; price and amount variables from 2017 to 2021 are adjusted based on the Shanxi provincial rural consumer price index (2017 = 100). A rural land contract management right certificate does not grant ownership of the land to the household; rather, it is a legal document that, following the activation of a rural land contract, the state recognizes as proof of the household’s right to contract and manage the land.

, participation in the program was treated as a dummy variable. Since the program was implemented in 2018, the year 2017 represented the time before the program implementation, and the year 2021 represented the time after the program implementation. If a farm household participated in the program, it was assigned 1 (the treatment group); if the sampled farm household did not participate, it was assigned 0 (the control group).

3.2.2. Dependent Variables

The dependent variables for this study were economic benefits, ecological benefits, and social benefits. Economic benefit indicators were panel data, including operational income, property income, and wages. Operational income refers to income generated from the program, including jujube forest operations, operations under jujube forests, and other agricultural and forestry activities. Property income pertains to income from the transfer of jujube forest lands, and wages refer to income from employment in jujube-related jobs. Ecological and social benefit indicators were cross-sectional data, measured through farm households’ perceptions with values assigned as follows: 1 = significant decrease, 2 = minor decrease, 3 = no change, 4 = minor increase, 5 = significant increase. Ecological benefit indicators included biodiversity, land resource conservation, and forest landscape, while social benefit indicators included employment opportunities and life satisfaction. These data were collected and calculated based on responses to the questionnaire (see Table 2).

3.2.3. Moderating Variables

To explore the mechanisms through which the JFHGOP influenced farmers’ welfare, this study constructed interaction terms between government subsidies and cooperative variables as moderating variables. The government subsidy variables included free provision of fertilizers and farming tools, equivalent to approximately 180 RMB per person; the cooperative variables indicate whether farmers have joined a cooperative.

3.2.4. Matching Variables

Referring to the existing research literature and considering data availability, the variables used in this study mainly included (1) individual characteristics: gender, age, level of education, and position, that is, whether they are village leaders [35]; (2) household characteristics: access to the internet, household size, labor force size and migrant labor force size [36,37]; (3) operational characteristics: arable land area and whether there is a rural land contractual management certificate [38,39]; (4) external environments: access to jujube technical training and subsidies for jujube forest cultivation [40,41].

Table 2 presents the descriptive statistics, and the test of initial sample mean differences for this study. The results indicated that the number of samples in the treatment group was roughly equivalent to that in the control group. As for the dependent variables, the treatment group showed significantly higher initial income from jujube cultivation and other agricultural and forestry income compared to the control group. As for the matching variables, the treatment group demonstrated higher levels across various indicators at the initial stage.

3.3. Estimation Methods

3.3.1. Propensity Score Matching (PSM)

The JFHGOP could be considered as a quasi-natural experiment. The double difference method (DID) is commonly used for evaluating such experiments. However, the household samples in Table 2 did not pass the parallel trend test, failing to meet the requirements of the DID method. Therefore, this study adopted the propensity score matching (PSM) method as an estimation method. This approach was chosen for two key reasons: First, since we cannot observe the counterfactual income of participating households had they not joined the program, directly comparing income differences between participants and non-participants could introduce endogeneity. Second, the FHGOP was implemented based on local government policy, making participation an exogenous variable. During the pilot phase, household participation was determined by the government rather than by household characteristics, meaning that participation was not random.

First, the Logit model was employed to estimate the conditional probability of households participating in the program, i.e., the propensity score (PS). The PS value was calculated as follows:

$$\mathsf{{\rm P}}\mathrm{S}={\mathrm{P}}_{\mathrm{r}}\left(\mathrm{C}=1|\mathrm{X}\right)=\mathrm{E}\left(\mathrm{C}=0\right|\mathrm{X})$$

(1)

where $\mathrm{C}=1$ indicates that the households participated in the program, $\mathrm{C}=0$ indicates that the households did not participate in the program, and X indicates the observable personal characteristics, household characteristics, operational characteristics, and external environments.

Next, the treatment and control groups were matched. We adopted five matching methods, followed by a balance test. Finally, the difference in economic benefits between the treatment group and control group households was calculated, that is, the average treatment effect on the treated (ATT), to assess the impact of the program on the economic benefits of households. ATT was calculated as follows:

$$\mathsf{A}\mathsf{{\rm T}}\mathsf{{\rm T}}=\mathsf{{\rm E}}\left({\mathsf{{\rm Y}}}_1|\complement =1\right)-\mathsf{{\rm E}}\left({\mathsf{{\rm Y}}}_0|\complement =1\right)=\mathsf{{\rm E}}({\mathsf{{\rm Y}}}_1-{\mathsf{{\rm Y}}}_0)| \complement =1$$

(2)

where ${\mathsf{{\rm Y}}}_1$ is the economic benefit and ${\mathsf{{\rm Y}}}_0$ is the economic benefit of participating households assuming they do not participate in the program. $\mathsf{{\rm E}}\left({\mathsf{{\rm Y}}}_1|\complement =1\right)$ is directly observable, but $\mathsf{{\rm E}}\left({\mathsf{{\rm Y}}}_0|\complement =1\right)$ is not directly observable and belongs to the counterfactual result, and we applied the propensity score matching method to construct the corresponding alternative indicators.

3.3.2. Ordinary Least Squares (OLS)

The impact of the program on households’ ecological and social benefits was analyzed through their perceptions. The data were cross-sectional and estimated using the ordinary least squares method (see below):

$${\mathrm{y}}_{\mathrm{i}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}}_{\mathrm{i}}+{\mathsf{\beta }}_2{\mathrm{x}}_{\mathrm{i}}+{\mathsf{\epsilon }}_{\mathrm{i}}$$

(3)

where ${\mathrm{y}}_{\mathrm{i}}$ represents the households’ perception of ecological and social benefits; ${\mathrm{t}\mathrm{r}\mathrm{e}\mathrm{a}\mathrm{t}\mathrm{e}\mathrm{d}}_{\mathrm{i}}$ denotes whether households participate in the program, with i = 1 for yes and i = 0 for no; ${\mathrm{x}}_{\mathrm{i}}$ is a control variable, and a township dummy variable was introduced to control regional differences. ${\mathsf{\epsilon }}_{\mathrm{i}}$ is the residual term.

4. Results

4.1. Baseline Regression Results and Analysis

4.1.1. PSM Results

To match the treatment and control group samples, a regression analysis was conducted on the conditional probability fitted values of the treatment group households. The maximum likelihood estimates from the Logit model are shown in

Table 3. Logit model estimation results of households’ participation in the program.

Variable Name	Estimated Coefficient	Standard Error	Z-Value
Gender	0.567	0.184	3.08 ***
Age	0.249	0.576	0.43
Education	0.226	0.113	2.00 **
Leader	0.113	0.526	0.21
Internet	0.430	0.194	2.21 **
Household size	−0.144	0.254	−0.57
Labor force size	0.755	0.238	3.18 ***
Migrant labor force size	0.088	0.503	0.17
Arable land area	−0.594	0.503	−0.17
Rural land contract management right certificate	1.011	0.4342	2.34 **
Jujube technical training	1.233	0.262	4.70 ***
Jujube forests operating subsidy	0.170	0.062	2.76 ***
constant term	−2.959	2.411	−1.23
LR ch2(12)		88.69 ***
Pseudo R2		0.107
Sample size		604

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively.

. To control for heteroscedasticity, continuous variables were logarithmically transformed, and the transformed data were then used for econometric regression.

Factors such as gender, education, internet access, labor force size, rural land contract management certificates, jujube technical training, and subsidies had a significant positive impact on household participation in the program, while other variables showed no effect (Table 3).

To ensure robustness, we employed five different matching methods: k-nearest neighbor matching (n = 4), caliper matching (caliper = 0.03), 1:4 caliper matching, kernel matching (bandwidth = 0.06), and local linear regression matching (bandwidth = 0.8). These parameters were selected after extensive testing, as they provided the optimal number of matched samples and the highest level of support.

Table 4. Mean treatment effects of program impacts on households’ economic performance.

Variable Name	Matching Method	Treatment Group Mean	Control Group Mean	ATT	t-Value
Jujube forest operations income	K-nearest neighbor matching	2.263	0.978	1.285	4.61 ***
	Caliper matching	2.263	1.155	1.108	4.03 ***
	1-to-4 match in Calipers	2.263	1.089	1.174	4.13 ***
	Nuclear matching	2.263	1.118	1.145	4.30 ***
	Local linear Regression Matching	2.263	1.117	1.146	3.39 ***
	Average value	2.263	1.091	1.172
Jujube forest understory operations income	K-nearest neighbor matching	1.342	0.579	0.764	3.31 ***
	Caliper matching	1.342	0.564	0.778	3.45 ***
	1-to-4 match in Calipers	1.342	0.579	0.764	3.25 ***
	Nuclear matching	1.342	0.605	0.737	3.37 ***
	Local linear Regression Matching	1.342	0.617	0.725	2.68 ***
	Average value	1.342	0.589	0.754
Other agricultural and forestry operations Income	K-nearest neighbor matching	4.090	3.016	1.074	2.85 ***
	Caliper matching	4.090	2.843	1.247	3.41 ***
	1-to-4 match in Calipers	4.090	2.892	1.197	3.11 ***
	Nuclear matching	4.090	2.879	1.211	3.43 ***
	Local linear Regression Matching	4.090	2.828	1.261	2.72 ***
	Average value	4.090	2.892	1.198
Jujube forests land transfer income	K-nearest neighbor matching	2.102	0.415	1.687	8.500 ***
	Caliper matching	2.102	0.277	1.825	9.37 ***
	1-to-4 match in Calipers	2.102	0.317	1.785	8.88 ***
	Nuclear matching	2.102	0.322	1.780	9.30 ***
	Local linear Regression Matching	2.102	0.326	1.775	7.19 ***
	Average value	2.102	0.331	1.770
Jujube employment income	K-nearest neighbor matching	0.260	0.532	−0.272	−1.76 *
	Caliper matching	0.260	0.504	−0.244	−1.67 *
	1-to-4 match in Calipers	0.260	0.504	−0.244	−1.67 *
	Nuclear matching	0.260	0.501	−0.240	−1.70 *
	Local linear Regression Matching	0.260	0.492	−0.231	−1.03
	Average value	0.260	0.507	−0.246

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. The growth rate was calculated as follows: growth rate = ATT/control mean × 100%.

provides detailed results. Household participation significantly increased income from jujube forest operations, jujube forest understory operations, other agricultural and forestry operations, and jujube forest land transfers at the 1% significance level. However, jujube employment income showed a significant decrease at the 10% level. On average, income from jujube forest operations, jujube forest understory operations, other agricultural and forestry operations, and jujube forest land transfers for participating households increased to 2.263, 1.342, 4.090, and 2.102, respectively, with growth rates of 107.35%, 127.99%, 41.430%, and 534.218%. Conversely, jujube employment income de-creased to 0.260.

The JFHGOP had a notably positive impact on the economic benefits of households, particularly in boosting income from jujube land transfers. This effect encouraged the scale management of jujube forests and heightened production enthusiasm, fostering the development of understory jujube forest operations, other agricultural and forestry operations, and forest land transfers. Consequently, Hypothesis 1 was validated.

4.1.2. Common Support Domains and Equilibrium Test

The five matching methods showed that the sample losses were 16, accounting for a relatively low proportion, indicating better sample matching. To more intuitively present the common support domain of the treatment and control groups, we used the k-nearest-neighbor matching of income from jujube forest operations as an example. The distribution of the propensity scores and the region of common support were shown in Figure 2. The figure illustrated the bias in the distribution of propensity scores between the groups of participants and non-participants, clearly highlighting the significance of proper matching and the application of the common support condition to avoid poor matches.

The use of the PSM method relies on another critical assumption: the balance test. This test requires that there be no systematic differences between the treatment and control groups across various matching variables after matching.

Table 5. Balance test for explanatory variables before and after matching.

Matching Method	Pseudo R2	LR Statistic	p-Value	Mean Bias	Med Bias
Prematch	0.108	90.30	0.000	25.6	27.0
K-nearest neighbor Matching	0.010	8.28	0.763	6.3	6.2
Caliper matching	0.010	8.19	0.770	6.2	6.5
1-to-4 match in calipers	0.011	9.47	0.663	7.3	7.6
Nuclear matching	0.008	7.22	0.843	5.7	6.7
Local linear regression Matching	0.018	15.86	0.198	9.5	9.8

presents the results of the balance test for explanatory variables before and after matching. Following the matching process, all parameters showed a decrease, and the p-values of likelihood ratio tests indicated that the mean differences in covariates between the paired samples were no longer significant. Therefore, the total bias in the samples significantly decreased after matching, indicating that the balance test was successful.

4.1.3. Least Squares Regression Estimation Results

The regression results for ecological and social benefits are presented in

Table 6. Regression results for ecological and social benefits.

Variable Name	Ecological Benefit			Social Benefit
	Biodiversity	Land Resource Conservation	Forest Landscape	Employment Opportunities	Life Satisfaction
Involvement in programs	0.131 (0.125)	0.243 * (0.127)	0.289 *** (0.103)	0.288 ** (0.129)	0.300 ** (0.116)
Control variable	Yes	Yes	Yes	Yes	Yes
Regional variables	Yes	Yes	Yes	Yes	Yes
Observed value	302	302	302	302	302
Constant term	3.912 ***	0.291 ***	4.571 ***	4.525 ***	1.498
Adjusted R2	0.081	0.170	0.111	0.096	0.114

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. Standard errors in parentheses.

. Regarding ecological benefits, participating households showed a significant increase in the perception of saving land resources (p < 0.1) and forest landscape (p < 0.01). In terms of social benefits, participating households perceived a significant improvement in job opportunities and life satisfaction (p < 0.05). Overall, the JFHGOP promoted both ecological and social benefits for rural households, thereby supporting Hypothesis 2.

4.1.4. Robustness Tests

(1)

Panel Tobit Model

Due to the presence of many zero values in the economic benefit indicators described above (characteristics of truncated data), we employed a panel Tobit model to further examine the impact of the JFHGOP on households’ economic benefits. The regression results are presented in

Table 7. Robustness tests based on panel Tobit models.

Variables	Jujube Forests and Operations Income (1)	Jujube Forests Understory Operations Income (2)	Other Agriculture and Forestry Income (3)	Jujube Forests Land Transfer Income (4)	Jujube Employment Income (5)
Involvement in programs	3.357 *** (0.853)	6.055 *** (1.886)	2.051 ** (0.889)	8.647 *** (0.676)	−5.309 ** (2.412)
Control variables	Yes	Yes	Yes	Yes	Yes
Time variables	Yes	Yes	Yes	Yes	Yes
Observed value	604	604	604	604	604
Wald chi2	149.88	35.27	36.19
Pseudo-R2				0.335	0.260

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. Columns (1) to (3) employ random effects panel Tobit regression, with standard errors in parentheses. Columns (4) and (5) use mixed Tobit regression, with clustered robust standard errors in parentheses.

. Even using different regression models, the economic benefits of households still showed a significant improvement, which was consistent with the baseline regression results.

(2)

Multiple Ordered Logistic Regression Models

Participating households of ecological and social benefits were characterized with ordered categorical variables and were suitable for ordered logistic regression analysis. The regression results were consistent with the baseline regression results (

Table 8. Robustness test of ecological and social benefits based on multivariate logistic regression models.

Variables	Ecological Benefits			Social Benefits
	Biodiversity	Land Resource Conservation	Forest Landscape	Employment Opportunities	Life Satisfaction
Involvement in programs	0321 (0.326)	0.553 * (0.288)	0.810 *** (0.308)	0.764 *** (0.309)	0.770 ** (0.304)
Control variable	Yes	Yes	Yes	Yes	Yes
Regional variables	Yes	Yes	Yes	Yes	Yes
Observed value	302	302	302	302	302
Pseudo R2	0.025	0.066	0.058	0.052	0.054

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. Data in parentheses are robust standard errors.

).

4.2. Heterogeneity Analysis

To further analyze the impact of the JFHGOP on farm household welfare under different resource endowments, we categorized households into groups based on education level, number of laborers, jujube forest planting scale, and annual household income. Following Li Xiaojing’s approach [36], we first calculated the mean values of the grouping variables and then compared the samples of households with “above-mean” and “below-mean” values. The economic benefits after grouping were estimated using the k-nearest neighbor matching method (

Table 9. Heterogeneity analysis of the economic benefits of the JFHGOP for households.

Grouping Variable		Jujube Forest Operations Income	Jujube Forest Understory Operations Income	Other Agriculture and Forestry Income	Jujube Forest Land Transfer Income	Jujube Employment Incomes
		ATT	ATT	ATT	ATT	ATT
Education level	Above mean	0.864 ** (2.31)	0.862 ** (2.18)	1.156 ** (2.19)	1.931 *** (6.55)	−0.162 (−1.12)
	Below mean	1.072 *** (2.71)	0.639 ** (2.03)	1.369 *** (2.77)	1.758 *** (6.48)	−0.283 (−1.27)
Labor force size	Above mean	1.587 *** (2.62)	0.870 (1.30)	2.410 *** (2.99)	1.591 *** (3.75)
	Below mean	1.085 *** (3.45)	0.724 *** (3.27)	0.548 (1.34)	1.730 *** (7.66)	−0.299 * (−1.65)
Average size of jujube forest plots	Above mean	0.751 (1.25)	1.105 * (1.86)		1.393 *** (3.82)	−0.676 (−1.62)
	Below mean	1.223 *** (4.06)	0.649 *** (2.98)		1.939 *** (7.70)	−0.090 (−0.64)
Annual household income	Above mean	1.619 *** (3.93)	1.135 *** (2.70)	1.284 ** (2.01)	1.969 *** (6.74)	−1.160 (−0.67)
	Below mean	0.730 * (1.88)	0.524 ** (2.07)	0.306 (0.65)	1.519 *** (5.75)	−0.181 (−0.86)

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. Values in parentheses are t-values.

), while the ecological and social benefits after grouping were estimated using the least squares method (

Table 10. Heterogeneity analysis of the ecological and social benefits of the JFHGOP for households.

Grouping Variable		Land Resource Conservation	Forest Landscape	Employment Opportunities	Life Satisfaction
Education level	Above mean	0.034 (0.219)	0.314 ** (0.151)	0.468 ** (0.220)	0.208 (0.183)
	Below mean	0.339 ** (0.147)	0.313 ** (0.141)	0.188 (0.166)	0.393 ** (0.162)
Labor force size	Above mean	0.548 * (0.312)	0.318 (0.229)	0.677 ** (0.290)	0.468 ** (0.194)
	Below mean	0.223 (0.139)	0.273 ** (0.122)	0.178 (0.153)	0.203 (0.146)
Average size of jujube forest plots	Above mean	0.062 (0.256)	0.081 (0.204)	−0.254 (0.223)	0.272 (0.244)
	Below mean	0.345 ** (0.157)	0.362 *** (0.121)	0.462 *** (0.160)	0.331 ** (0.144)
Annual household income	Above median	0.325 * (0.197)	0.307 * (0.166)	0.490 ** (0.204)	0.530 *** (0.170)
	Below median	0.117 (0.162)	0.287 ** (0.131)	−0.010 (0.167)	0.066 (0.157)

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. Values are robust standard errors.

).

4.2.1. Heterogeneity Analysis of Economic Benefits

Relatively speaking, participating households with lower education levels and larger labor forces experienced significant increases in operating income (Table 9). One possible reason for this was that households with lower education levels among their labor force relied more heavily on agricultural and forestry income. In contrast, households with larger labor forces had sufficient human capital to invest in production and operations. This allowed them to participate in the program, increase yields, and achieve higher income.

It is notable that the effect of the program on participating households with larger labor forces and those obtaining jujube forest understory operating income was not significant; however, the effect on households with smaller labor forces was significant. One possible reason for this difference was the rapid aging of the local labor force. Households with smaller labor forces participated in jujube forest management and understory management but lacked the capacity to undertake other agricultural and forestry production and operations. In contrast, households with larger labor forces possessed higher human capital and achieved higher levels of agricultural and forestry output, regardless of their participation in the program.

Additionally, participating households with higher education levels and larger labor forces saw significant increases in property income, indicating that households with higher human capital had more access to information and revenue channels, resulting in a smaller proportion of their income coming from operating activities. They were able to earn income through forest land transfers.

We divided the participating households into different groups based on the average size of their jujube forest plots. Households with smaller jujube forest plot sizes experienced significant increases in both operational and property incomes due to their participation in the program (Table 9). In contrast, the income effects for households with larger plot sizes were not significant. This indicated that the program effectively alleviated constraints on production inputs and operations for smaller-scale farmers, leading to income growth. It should be noted that “a large country with small-scale farm households” remains a fundamental characteristic of China’s national context. Similarly, small-scale households continue to play a dominant role in jujube production and management.

According to the median annual income, the program had a greater positive impact on the economic benefits of participating households with higher annual incomes (Table 9). This was because households with higher incomes had stronger resource endowments, allowing for better alignment between technological elements and other factors, which led to higher output.

4.2.2. Heterogeneity Analysis of Ecological and Social Benefits

In terms of ecological benefits, participating households with lower resource endowments perceived significant positive effects (Table 10). Specifically, households with lower education levels and smaller jujube forest plot sizes observed noticeable improvements in the surrounding ecological environment because of their participation in production and operations. Both households with higher and lower annual incomes reported significant positive effects on forest landscapes, indicating that the program had a substantial impact on enhancing the local forest ecological environment.

In terms of social benefits, households with a larger labor force perceived significant positive effects, while those with higher education levels reported only a notable improvement in employment opportunities. Conversely, households with lower education levels experienced a more substantial increase in happiness. Additionally, households with smaller jujube forest plot sizes and higher annual incomes also noticed more pronounced social benefits.

4.3. Analysis of the Pathway

The above results indicated that the JFHGOP enhanced the economic, ecological, and social welfare of farm households in the LM area. To explore the pathways through which it influenced farm household welfare, we employed moderation effect analysis, focusing on two pathways: government subsidies and support from cooperatives. Using cross-sectional data, we applied Formula 3 as the regression model, with the dependent variables being the economic, ecological, and social benefits of households. Interaction terms for the moderation variables were included in the model during the analysis.

Government subsidies played a regulatory role in the welfare of households participating in the JFHGOP (

Table 11. Analysis of the moderation mechanisms of government subsidies on economic, ecological and social benefits.

Variable Name	Jujube Forest Operations Income	Jujube Forests Understory Operation Income	Other Agriculture and Forestry Income	Jujube Forest Land Transfer Income	Jujube Employment Incomes	Land Resource Conservation	Forest Landscape	Employment Opportunities	Life Satisfaction
Program	−0.04 (−0.10)	0.26 (0.28)	2.06 *** (0.58)	5.52 *** (0.23)	−2.80 × 10−14 (3.98 × 10−9)	0.38 ** (0.15)	0.26 *** (0.09)	0.19 (0.15)	0.29 ** (0.13)
Program × subsidies	0.53 *** (0.10)	0.56 *** (0.12)	0.36 ** (0.17)	−1.01 *** (0.06)	0.06 *** (0.02)	0.32 *** (0.04)	0.14 *** (0.02)	−0.01 (0.04)	0.27 *** (0.34)
Subsidies	−0.47 (0.11)	−0.51 *** (0.16)	−0.51 *** (0.19)	0.61 *** (0.11)	0.03 *** (0.10)	−0.34 *** (−0.04)	−0.20 *** (0.01)	0.06 * (0.04)	−0.33 *** (0.04)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Regional variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observed value	302	302	302	302	302	302	302	302	302
R2	0.24	0.15	0.05	0.72	0.27	0.09	0.09	0.05	0.07

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. Standard errors are presented parentheses.

). In terms of economic benefits, the interaction between government subsidies and participating households had a significant positive impact on operating income. However, it had a significantly negative impact on income from jujube forest land transfers and jujube employment. This indicated that government subsidies stimulated households’ enthusiasm for production and operations. Furthermore, our investigation found that households managing jujube forests were more likely to receive government subsidies. In terms of ecological and social benefits, households receiving government subsidies perceived a significant increase in land resource conservation, forest landscapes, and overall happiness. This indicated that under government subsidies, households were more satisfied with the environmental changes due to jujube forest management, and their perception of ecological benefits was positive. However, the impact on employment opportunities was not significant, likely because households receiving government subsidies were more inclined to participate in jujube forest management, coupled with the influence of aging that reduces employment opportunities.

In summary, government subsidies improved the level of jujube forest management and resulted in higher economic and ecological benefits and increased life satisfaction, thereby supporting Hypothesis 3.

The moderation analysis of cooperatives’ impact on the welfare of participating households is detailed in

Table 12. Analysis of the regulatory mechanisms of cooperatives for economic, ecological and social benefits.

Variable Name	Jujube Forest Operations Income	Jujube Forests Understory Operation Income	Other Agriculture and Forestry Income	Jujube Forest Land Transfer Income	Jujube Employment Income	Land Resource Conservation	Forest Landscape	Employment Opportunities	Life Satisfaction
Program × Cooperatives	−0.11 (0.16)	0.52 * (0.27)	2.08 *** (0.45)	4.87 *** (0.25)	0.80 *** (0.21)	0.37 *** (0.12)	0.24 *** (0.09)	0.33 *** (0.12)	0.12 ** (0.10)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Regional variables	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Observed value	302	302	302	302	302	302	302	302	302
R2	0.24	0.15	0.07	0.71	0.22	0.08	0.05	0.05	0.04

Note: *, **, *** represent the significance levels of 10%, 5%, and 1% respectively. Due to multicollinearity between the interaction terms of the program and cooperatives and the main interaction terms, all of which are categorical variables, only the interaction terms are used for regression. Standard errors are presented in parentheses.

. Economically, cooperatives significantly influenced income from other agriculture and forestry activities, jujube forest land transfers, and jujube-related employment. This suggests that cooperatives boosted households’ property income through better land resource integration and increased wage income from cooperative work. In terms of ecological and social benefits, households participating in cooperatives reported notable improvements in both areas.

Cooperatives increased households’ awareness of ecological benefits and improved their perceptions through ecological education and outreach. Additionally, by offering local employment opportunities, cooperatives enhanced households’ sense of belonging and satisfaction, leading to greater overall happiness, thereby supporting Hypothesis 4.

5. Discussion

Previous studies have primarily focused on the ecological benefits of jujube forests, such as soil quality and water retention [1,42,43]. In contrast, this study explored these benefits from the perspective of farm household welfare. The results indicated a significant improvement in households’ perceptions of land resource conservation and forest landscape enhancement. Additionally, the adoption of new varieties generally has a positive impact on farm household welfare [44,45], which is consistent with our findings. For instance, introducing new legume varieties in Tanzania and Ethiopia has increased employment, absorbed surplus labor, and expanded income sources [45]. Data from three major grain-producing provinces in China indicate that new maize varieties increase yields by 9.4% and mitigate the negative effects of natural disasters [46]. In Ghana, adopting improved tomato varieties has led to higher household expenditure and asset accumulation [47]. These studies suggest that variety improvements are cost-effective and sustainable ways to enhance food security, raise incomes, and reduce poverty in developing countries [48,49,50]. This study, however, offers a comprehensive analysis of how forestry technological innovation impacts the economic, ecological, and social welfare of farm households.

Although the JFHGOP yielded positive outcomes, this study found that it did not increase jujube employment income for households. This may be related to the global trend of an aging agricultural workforce [51,52]. The survey revealed that this was significant in the LM area, where the physical demands of jujube production are challenging for elderly workers [52] or result in lower productivity [53]. Consequently, jujube-related employment income did not increase. While the program did not boost jujube employment income for the households surveyed, it created more job opportunities in jujube production for younger laborers, thus enhancing the regional social benefits. Additionally, the program may have encouraged farm households to focus more on jujube forest management, reducing the demand for jujube-related employment.

Research by Shen H. et al. also indicates that households with lower education levels struggle to secure good off-farm jobs and rely more on land-based income. Households with larger labor forces, dependent on agriculture and forestry, also show a greater reliance on land [5]. These findings are consistent with this study, indicating that education level and labor force size are key factors influencing farm household welfare in the LM area.

It is also worth noting that some studies suggest that smallholders tend to benefit more from agricultural and forestry technological innovations in terms of household welfare, which is consistent with our findings [36,37,45,46,54]. This may be because large-scale farmers typically have greater resource endowments before joining technology promotion programs, giving them better access to variety information and achieving higher yields and income. In contrast, smallholders may experience more significant gains from these programs. Therefore, our study further suggests that smallholders with fewer resources need stronger policy support.

This study confirms the reinforcing effects of government and cooperative support on the JFHGOP. Proactive forestry policies are known to boost household income [55,56], and government support, including subsidies, can encourage smallholders’ entrepreneurial efforts [57,58], lower jujube forest input costs, and stimulate household production and management. These findings align with this study’s results. For example, in Shibaitou Township’s Baizhai Village, the Qiao family saw a 300.6% return on jujube operations and a 75.75% return from jujube forest understory operations after receiving government subsidies. In Shibaitou Township’s Gaojiaju Village, the Gao family participants achieved a 9500% return on jujube cultivation and a 1150% return from understory operations.

However, not all government subsidies are equally effective. Lu et al. found that afforestation and nurturing subsidies did not significantly impact household income, while ecological subsidies had a positive effect. This may be due to the clear and effective implementation of ecological subsidy standards, which are more strongly perceived by households [56]. This underscores the importance of maintaining and enhancing the stability and effectiveness of government subsidy policies in practice.

In addition to subsidies, cooperatives significantly impact farm household welfare [59,60,61]. Research by Ma and Abdulai, for instance, found that cooperative members experienced notable increases in apple production and household income [60]. Our study supports these findings, indicating that cooperatives enhance the JFHGOP’s effectiveness, leading to higher income for households. In this study, the per capita income from other agriculture and forestry for households participating in the cooperative was 4132.803 CNY, compared to 1197.015 CNY for non-participating households. The former is significantly higher, with a difference of 2935.788 CNY and a percentage difference of approximately 245.8%. Thus, support from cooperatives proves to be an effective means of improving farm household welfare.

Our study further reveals that the scale of cooperative operations positively influences farmers’ property income. In this study, land transfers through cooperatives offered payment rates of 4200 CNY and 7500 CNY per hectare. Among participating households, the highest earnings reached 90,903.78 CNY, while non-participating households did not receive any income from this source. It is important to note that this impact largely depends on cooperatives that lead agricultural production and possess substantial financial resources [44]. In Lüliang, local governments actively support professional cooperatives and other jujube-related businesses [5], creating a conducive environment for the sustainable development of jujube forests. However, it is necessary to further explore whether cooperatives can continue to enhance farm household welfare in the absence of government support.

Meanwhile, the research by Mojo et al. demonstrates the significant role of cooperatives in both ecological and social aspects [62,63,64]. Our findings suggest that cooperatives can effectively facilitate program promotion due to their training functions [65]. Since households are already familiar with jujube cultivation, cooperatives can more easily advance the program, increasing households’ awareness of jujube forest management and changes in forest landscapes. Additionally, by providing local employment opportunities, cooperatives help prevent farmers from migrating away from their families, thereby enhancing their overall well-being.

6. Conclusions

(1)

The Jujube Forest High Grafting and Optimization Program significantly impacts households’ overall welfare, including economic, ecological, and social benefits. These conclusions hold firm even after passing robustness tests.

(2)

Heterogeneity analysis reveals that households’ resource endowments are critical for improving welfare. Factors such as education level, labor force size, jujube forest plot size, and annual household income each play a significant role.

(3)

Path analysis shows that the program influences farm household welfare through government subsidies and cooperative support.

Based on these findings, we propose the following policy recommendations:

(1)

To achieve a balance between economic development and ecological protection in the middle reaches of the Yellow River, it is essential to advance forestry technological innovation and consolidate jujube forest areas.

(2)

Smaller farm households remain the backbone of jujube industry production and management. Therefore, governments should offer additional support through policies, funding, and technology.

(3)

Attention should also be given to labor force aging and migration in impoverished and remote mountainous areas. Providing forestry production and labor replacement services can support sustainable development in mountainous forestry and improve farm household welfare.

(4)

Emphasizing and optimizing subsidy mechanisms in remote mountainous areas is crucial. Additionally, industry assistance policies should be further implemented to supplement production factors. This will help fully leverage policy subsidies in supporting the production and management of smaller farm households.

(5)

Finally, further support cooperative organizations and other new business entities, cultivate leading enterprises, and promote collaboration among the government, these new entities, and farm households. This will advance green development in mountainous forestry and enhance farm household welfare.