Impact of Aging Agricultural Labor Force on Mulch Film Recycling Behavior: Evidence from Rural China

Abstract

Mulch film recycling is essential for reducing soil pollution, ensuring sustainable land use, and promoting green agricultural development. This study examines the impact of agricultural labor age on mulch film recycling behavior, utilizing survey data from 739 households in Xinjiang. The relationship between labor age and recycling behavior follows an inverted U-shape, with participation initially increasing and then decreasing as age advances. We explore the mediating roles of human capital and ecological cognition in this relationship. Both human capital and ecological cognition display inverted U-shaped patterns with age, which in turn influence recycling behavior. Furthermore, social norms were found to positively moderate the relationship between labor age and ecological cognition, while no significant moderating effect was observed between age and human capital. These findings suggest that enhancing human capital and ecological awareness, coupled with the reinforcement of social norms, can facilitate mulch film recycling. The study underscores the importance of developing targeted policies to support various age groups in adopting sustainable agricultural practices.

1. Introduction

The aging of agricultural labor has become a critical constraint in advancing global sustainable agricultural practices, with China’s Xinjiang region serving as a paradigmatic case. In China, the intensive agricultural reliance on plastic film has resulted in over 200 million tons of plastic residues annually, contaminating farmland spanning 2 million hectares [1]. Xinjiang has both achievements and challenges in the treatment of agricultural residual plastic film. The three-tier governance system at provincial, municipal, and county levels, supported by subsidy policies and mechanized recovery technologies, has achieved a seasonal recovery rate of 80%. However, this progress masks emerging contradictions: elderly farmers, who constitute the backbone of plastic film recovery efforts, face physical limitations and slow technological adoption rates that hinder the effective implementation of policies. This dilemma reflects broader agricultural transformation trends, as rural youth migration intensifies and global agriculture must balance production and environmental demands with aging populations. Investigating the mechanisms through which labor aging influences plastic film recovery behaviors holds significant implications for resolving the global synergy challenges between agricultural sustainability and demographic transitions.

Scholars have conducted extensive research on the relationship between the age of agricultural laborers and their green production behaviors. Some studies suggest that as farmers age, declines in physical strength, learning ability, and land transfer [2,3], combined with shorter ecological benefit periods compared to younger individuals [4] have inhibited farmers ‘participation in green production behavior. However, other studies have reached opposite conclusions, arguing that as farmers age, their attachment to the land increases [5], significantly enhancing their adoption of green production behaviors. Additionally, some studies suggest that the relationship between age and farmers’ green production behaviors is not linear but follows an inverted U-shape [6,7]. This may reflect a shift in farmers’ production goals with age—from treating farming as a supplementary livelihood to relying on it as the main source of income, and eventually to subsistence-oriented farming—as physical strength, opportunity costs, and livelihood dependence change over time [8].

While the existing research has achieved rich conclusions in related fields, which lays a foundation for the analysis of this paper, there are still the following aspects to be further discussed. First, existing literature primarily investigates the linear relationship between age and farmers’ green production behaviors, with limited exploration of their nonlinear relationship. Second, some scholars believe that human capital and risk preferences are mediating variables between farmers’ age and green production behavior [8], but few studies explore the mediating role from the perspectives of ecological cognition and human capital. Third, existing literature has found that social network embedding can mitigate the negative impact of aging on farmers’ agricultural technology extension services [9]. “Social network embedding” refers to the extent to which farmers are connected and interact within local social networks, including kinship, neighborhood, and cooperative relationships. This interaction can mitigate the negative impact of aging on participation in agricultural technology extension services [9]. Through these embedded social ties, information, labor, and behavioral norms can be transmitted, influencing individual decision-making and production behavior. In the context of rural China, “agricultural technology extension services” mainly refer to governmental and collective mechanisms that provide farmers with access to new agricultural technologies, training, and information. These services are usually delivered through local agricultural bureaus, cooperatives, or agricultural technicians, and their effectiveness depends heavily on farmers’ social network connections and participation. However, few studies have examined the moderating role of social norms in the relationship between age, human capital and ecological cognition. Therefore, this paper takes farmers’ mulch film recycling behavior as an example to explore the relationship between labor age and their mulch film recycling behaviors. It constructs a model to test the mediating effects of human capital and ecological cognition in this relationship and further explores the moderating role of social norms. The goal is to deepen the understanding of how labor age affects farmers’ green production behaviors and provide decision-making references for improving the agricultural ecological environment and achieving sustainable agricultural development.

2. Theoretical Framework and Research Assumptions

2.1. Labor Age-Human Capital-Mulch Film Recycling Behavior

Human capital consists of economically valuable knowledge, skills, abilities, and health qualities aggregated in workers, reflecting their quality [10]. With increasing age, farmers’ production skills continuously improve, demonstrating an “accumulation effect” of experience. However, their physical strength and health decline, showing an “aging effect” on physical ability. The change in labor productivity depends on the relative strength between these two effects of experience accumulation and physical decline. If the former outweighs the latter, labor supply capacity increases; otherwise, it decreases [9]. Specifically, during their youth, the “accumulation effect” plays a predominant role, with human capital gradually increasing with age until reaching its peak in middle age. In their old age, the “aging effect” becomes predominant, leading to a decline in human capital with further aging. Meanwhile, human capital is one of the most critical resources for farmers’ production activities. Under unchanged conditions, accumulated human capital benefits agricultural productivity. Farmers with higher human capital possess greater knowledge reserves, healthier physical conditions, and stronger production capabilities [11], thereby significantly promoting their environmentally friendly behaviors [12,13], such as mulch film recycling. Based on the above analysis, the study proposes a hypothesis:

Hypothesis 1 (H1).

There is an inverted U-shaped relationship between age and human capital.

Hypothesis 2 (H2).

Human capital mediates the inverted U-shaped relationship between age and farmers’ mulch film recycling behavior.

2.2. Labor Age-Ecological Cognition-Mulch Film Recycling Behavior

Ecological cognition refers to individuals’ basic understanding of the surrounding ecological environment from the perspective of ecological services, reserves of ecological scientific knowledge, and the formation of ecological awareness, among other aspects. Farmers in different age groups belong to different generational cohorts, and their actual differences are the result of the combined effects of age effects and generational effects [14]. Age effects refer to the differentiation in individuals’ psychological characteristics as they go through life stages [15]. Due to their different life stages, farmers in different age groups have varying roles and levels of interaction with agricultural land [16]. Generational effects reflect behavioral and cognitive differentiation among different generational cohorts, in which farmers from different age groups differ in their acceptance of new technologies, knowledge of agricultural innovations, and understanding of sustainable development concepts [17,18]. Given the constraints and decision-making goals faced by farmers in different age groups, the combined effects of these two factors on ecological cognition vary with age. Considering the current rural situation, young farmers primarily rely on non-agricultural income, which somewhat limits their ecological cognition and their inclination to participate in mulch film recycling behavior. Middle-aged farmers with limited non-agricultural employment opportunities spend more time in agricultural production, accumulating richer agricultural production experience [19]. They understand the importance of mulch film recycling in increasing yields and improving land quality and are willing to participate. Elderly farmers are nearing retirement [16], have shorter periods of ecological benefits from mulch film recycling, and thus show lower attention to ecological cognition, reducing their likelihood of participating in mulch film recycling. Based on the above analysis, the study proposes a hypothesis:

Hypothesis 3 (H3).

There is an inverted U-shaped relationship between age and ecological cognition.

Hypothesis 4 (H4).

Ecological cognition mediates the inverted U-shaped relationship between age and farmers’ mulch film recycling behavior.

2.3. The Regulatory Effect of Social Norms

Social norms refer to the widely accepted rules or standards of behavior that members of a group follow [20]. These norms are transmitted and reinforced through social interaction, group pressure, or customs. Rural society consists of networks of farmers and their mutual relationships [21,22]. Social norms rooted in rural communities continuously expand the added value of individual human capital, compensating for the depreciation of human capital and enhancing farmers’ own endowment of human capital [23], which facilitates their participation in plastic film recovery. Specifically, social norms broaden farmers’ human capital through at least two mechanisms. First, the support mechanism of mutual labor. The adoption of green production behaviors by farmers depends to some extent on the endowment of labor resources or the labor cost of green production behaviors [24,25,26]. Through cooperative and reciprocal relationships, networks of relatives and friends provide mutual support in labor, solving the problem of insufficient labor supply during plastic film recovery [27]. Second, the interactive learning mechanism. In rural community life, farmers accumulate agricultural production skills through mutual communication, imitation, and other means, compensating for the labor weakness of young laborers engaged in non-agricultural work and elderly laborers lacking physical strength [9]. Based on the above analysis, this study proposes a hypothesis:

Hypothesis 5 (H5).

Social norms have a positive regulatory effect on the inverted U-shaped relationship between age and human capital.

Social cognitive theory posits that the social environment plays a crucial role in shaping individual cognition, intentions, and behaviors [28]. As a unique social environment, social norms influence the ecological cognition and green production behaviors of farmers of different age groups. Specifically, social norms affect farmers’ ecological cognition through two mechanisms. First, internalization through interaction and value guidance. In the agricultural production process, most people tend to conform and take agricultural decision-making information conveyed by relatives and friends as the correct basis, internalizing the views of other farmers as their own behavioral standards [29,30]. Farmers often learn through social interaction that others believe plastic film should be recycled, leading them to rationally analyze and evaluate the benefits of recycling plastic film, correct cognitive biases towards plastic film recovery, and actively engage in plastic film recovery behaviors. Second, reducing information asymmetry. Social norms serve as an important platform for interactive learning among farmers, stabilizing and intensifying communication, thereby reducing farmers’ costs of understanding and participating in plastic film recovery and effectively enhancing their understanding of green production behaviors [10], significantly promoting their participation in plastic film recovery behaviors. Based on the above analysis, this study proposes a hypothesis:

Hypothesis 6 (H6).

Social norms have a positive regulatory effect on the inverted U-shaped relationship between age and ecological cognition.

Building on the previous hypotheses (H1–H6), which establish that human capital and ecological cognition evolve with age and affect farmers’ recycling behavior. Specifically, as age increases, farmers’ participation in recycling behavior initially rises due to the accumulation of human capital and enhanced ecological cognition. However, after a certain age, physical decline and diminishing ecological awareness reduce their likelihood of participating in recycling, leading to an inverted U-shaped relationship between age and recycling behavior. Based on the above analysis, this study proposes a hypothesis:

Hypothesis 7 (H7).

There is an inverted U-shaped relationship between age and farmers’ mulch film recycling behavior.

Accordingly, the research logic of this study is as follows: while the age of labor affects farmers’ plastic film recovery behaviors directly, it also has an indirect impact through two intermediate mechanisms—human capital and ecological cognition—and is moderated by social norms. Based on the research hypotheses, a diagram of the impact pathways between farmers’ age and their plastic film recovery behaviors has been constructed (Figure 1).

3. Materials and Methods

3.1. Study Area

3.1.1. Agricultural Production Characteristics of the Survey Area

Xinjiang, located in the northwest inland region of China, is characterized by long sunlight hours (2500–3500 h annually) and low annual precipitation (around 150 mm), making it a typical oasis irrigation agricultural area. The dry climate and low rainfall have driven the widespread use of plastic film mulching technology across Xinjiang, with over six regions having a mulching area exceeding 100,000 hm². The top five regions with the highest mulching coverage are Aksu, Kashgar, Bayingol, Changji, and Tacheng, where the overall mulching rate exceeds 60%. Below are the detailed agricultural production characteristics of the surveyed regions, demonstrating the rationality and representativeness of the data sampling locations.

Aksu Region: The region is characterized by a dry climate with high evaporation, minimal rainfall, long sunshine hours, and abundant heat resources. The cultivated area is about 660,000 hm², with a mulched area of about 620,000 hm². Major crops include cotton and wheat. Kashgar Region: Located in a warm temperate continental arid climate zone, Kashgar has distinct seasons, long sunshine hours, large temperature variations, and little rainfall. The cultivated area is about 690,000 hm², with a mulched area of about 540,000 hm². The main crops grown are cotton, wheat, and corn. Bayingol Mongolian Autonomous Prefecture (abbreviated as Bayingol): This area has arid conditions with little rainfall, large evaporation, and long sunshine hours. The high-altitude areas have a long spring and autumn but no summer, while the plains have distinct seasons. The cultivated area is around 360,000 hm², with a mulched area of approximately 330,000 hm², with cotton as the core crop (over 80% of the planted area). Changji Region: A typical continental dry climate, with more precipitation in the southern part during summer and desert climate features in the north. The cultivated area is approximately 740,000 hm², with a mulched area of about 290,000 hm². Main crops include cotton, wheat, and corn. Tacheng Region: Located in the temperate arid and semi-arid climate zone, the spring temperature increases quickly, and there are large temperature fluctuations. The cultivated area is around 790,000 hm², with a mulched area of approximately 410,000 hm², and the main crops are cotton and corn.

3.1.2. Rural Population Structure Characteristics in the Survey Area

The survey area, being the agricultural core of Xinjiang, exhibits a rural population structure characterized by significant aging, noticeable outflow of labor, and a high prevalence of part-time farming. The key characteristics are as follows:

High Overall Aging Level: According to the “Xinjiang Statistical Yearbook (2022)” and the national economic and social development statistics of various regions, the percentage of rural residents aged 60 and above in Aksu, Kashgar, Bayingol, Changji, and Tacheng is 18.2%, 17.5%, 19.3%, 16.8%, and 17.1%, respectively. Bayingol, located at the edge of the Tarim Basin, faces relatively harsh agricultural conditions and a significant outflow of young labor, resulting in the highest rural aging. Changji, being close to the Urumqi metropolitan area, has a lower aging level but still experiences seasonal migration of young labor.

Link Between Aging and Land Management Scale: In villages with higher aging levels (e.g., Ruoqiang County in Bayingol and Keping County in Aksu), land fragmentation is prevalent, with an average cultivated area per household of less than 0.3 hm². Elderly farmers struggle to perform mechanized mulch film recovery and mostly rely on manual collection or disposal. In contrast, larger-scale planting areas like Changji and Tacheng, with an average cultivated area of 1–2 hm² per household, have middle-aged and younger farmers who prefer managing their land through “land transfer + hired labor” models. In these areas, mulch film recovery is often entrusted to professional cooperatives or agricultural machinery service organizations, but elderly small-scale farmers still account for over 40% of the population, becoming a weak link in the management of film residues.

3.2. Data Sources

The data for this study were derived from a sampling survey of cotton growers conducted by the research team from July 2024 to August 2024 in Aksu, Kashgar, Bayingol, Changji Hui Autonomous Prefecture and Tacheng areas of Xinjiang Uygur Autonomous Region. Sample areas were selected based on principles of average distribution and representativeness. The sampling employed a stratified random sampling method, with 5 cities selected for the survey. In each city, 4 townships were randomly chosen, and from each township, 3 villages were randomly selected. Within each village, 10–15 households were chosen at random to participate in the survey. The research team conducted one-on-one questionnaire surveys and interviews with household heads or key family members involved in production decision-making. A total of 798 questionnaires were collected, and after excluding invalid questionnaires lacking key data or with logical inconsistencies, 739 questionnaires remained valid, yielding an effective response rate of 92.61%.

3.3. Research Methods

Since the dependent variable in this study, “whether the farmer recycled plastic film,” is a binary variable, a binary Logit regression model is used as the analytical model. This choice helps to avoid the issue of heteroscedasticity in the error term distribution that can arise from using a linear probability model based on Ordinary Least Squares (OLS) estimation, as well as the potential problem of the model’s predicted values exceeding the reasonable range for probabilities. The model is expressed as follows:

$$Logit\left(y_i=1\right)={\alpha }_0+{\alpha }_{1}Age+{\alpha }_2{Age}^2+{\gamma }_{i}X+{\epsilon }_i$$

(1)

In Equation (1), i represents the i-th farmer, and $Logit\left(y_i=1\right)$ represents the probability that the ith i-th farmer recycles plastic film. Age is the actual age of the agricultural decision-maker, and $X$ represents the control variables. ${\alpha }_0$ is the constant term, ${\alpha }_1$ and ${\alpha }_2$ are the regression coefficients of the key explanatory variables, ${\gamma }_i$ represents the total effect of the control variables, and $\epsilon$ is the random error term.

To examine the mechanism through which labor age affects farmers’ plastic film recycling behavior, this study tests from two dimensions: human capital and ecological cognition. Given the controversy surrounding the traditional “three-step approach” for testing mediation effects, this study adopts the “two-step method” approach suggested by Bostwick et al. (2022) [31] to construct the mediation effect model as follows:

$$M={\beta }_0+{\beta }_{1}Age+{\beta }_2{Age}^2+{\gamma }_{1}X+{\epsilon }_i$$

(2)

In this context, the mediator variable M represents either human capital or ecological cognition. ${\beta }_0$ is the constant term, ${\beta }_1$ and ${\beta }_2$ are the regression coefficients of the key explanatory variables, ${\gamma }_1$ represents the total effect of the control variables, and ${\epsilon }_i$ is the random error term.

Finally, the moderating effect of social norms is tested. The study examines the moderating role of social norms in the inverted U-shaped relationships between age and human capital, as well as age and ecological cognition.

$$\mathrm{H}\mathrm{u}\mathrm{m}={\delta }_{0 }+{\delta }_{1}Age+{\delta }_2{\left(Age\right)}^2+{\delta }_{3}SN+{\delta }_{4}SN\times Age+{\delta }_{5}SN\times {\left(Age\right)}^2+{\delta }_{6}X+{\mu }_1$$

(3)

$$\mathrm{E}\mathrm{c}\mathrm{o}={\beta }_{0 }+{\beta }_{1}Age+{\beta }_2{\left(Age\right)}^2+{\beta }_{3}SN+{\beta }_{4}SN\times Age+{\beta }_{5}SN\times {\left(Age\right)}^2+{\beta }_{6}X+{\mu }_{2 }$$

(4)

3.4. Variable Definitions and Measurement

Dependent Variable: The dependent variable is measured by the question, “Have you participated in plastic film recycling over the past year?” This captures the farmer’s recycling behavior during the most recent production year. The variable is coded as 1 if the farmer has participated and 0 otherwise.

Independent Variables: (1) The explanatory variable is age. Age is measured by the survey question: ‘What is your age (in years)?’. The age of the agricultural decision-maker is the core variable, based on theoretical analysis and research hypotheses. Age is examined to determine if there is an inverted U-shaped relationship between age and farmers’ plastic film recycling behavior. An additional variable, age squared, is included, and it is divided by 100 to improve the readability and accuracy of the model’s regression coefficients.

(2) Mediating Variables: Ecological cognition was measured by survey questions: “Perception of the severity of farmland plastic film pollution,” “Perception of environmental improvement resulting from plastic film recycling,” “Perception of yield enhancement due to plastic film recycling,” and “Perception of the importance of plastic film pollution control to farmers.” This study employed SPSS 27.0 and AMOS 26.0 to examine the reliability and validity of the measurement scales, and the results are presented in

Table 1. Reliability and validity test results.

Variable	Indicators	Factor Loadings	Cronbach’s α	CR	AVE	KMO
ECO	ECO1	0.802	0.764	0.850	0.586	0.688
	ECO2	0.758
	ECO3	0.748
	ECO4	0.752
SN	SN1	0.815	0.860	0.827	0.614	0.721
	SN2	0.811
	SN2	0.722

The questionnaire demonstrated good internal consistency, with a Cronbach’s Alpha coefficient of 0.764. The Kaiser–Meyer–Olkin (KMO) value was 0.688, indicating that the data were suitable for factor analysis. Principal component analysis with varimax rotation extracted a single common factor—labeled as the Ecological Cognition Factor—from the four observed variables. All item loadings exceeded 0.6. Furthermore, the composite reliability (CR) was greater than 0.8, and the average variance extracted (AVE) was above 0.5, providing evidence of satisfactory convergent validity for the measurement model.

Human capital reflects the quantity and quality of labor, determined by the education level of the labor force, the technical skills they possess, and their overall health status. In this study, human capital was measured by survey questions: “What is your highest level of education?” “How do you evaluate your health status?” and “How capable are you of recycling plastic film?”. The arithmetic mean of these three indicators was used as the observed value for human capital. Factor analysis was not conducted for this variable because these three indicators represent distinct yet complementary dimensions of human capital—namely, the cognitive dimension (education), the physiological dimension (health), and the practical skill dimension (recycling ability)—rather than a single homogeneous latent construct. This treatment aligns with Becker’s Human Capital Theory (Becker, 1964) [32], which conceptualizes human capital as encompassing education, health, and work-related skills that collectively enhance labor productivity but cannot be assumed to share a unified measurement dimension. Subsequent research has further emphasized the multidimensional and complementary nature of human capital components [10]. Consistent with prior studies on agricultural human capital measurement, the arithmetic mean of these three indicators was retained to represent the overall level of human capital. This approach reflects the theoretical understanding that education, health, and skills are mutually reinforcing yet non-substitutable elements that jointly determine an individual’s productive capacity.

(3) Moderating Variable: Social norms are measured by three questions: “Family members believe plastic film should be recycled,” “Neighbors believe plastic film should be recycled,” and “Village cadres believe plastic film should be recycled.” The results are presented in

Table 2. Descriptive statistics of variables.

Variable	Definition		Mean	Standard Deviation
Dependent variable
Mulch film recycling behavior	Recycle mulch film: (No = 0, Yes = 1)		0.520	0.500
Independent variable
Age	Age	Actual age/years of the head of household	47.880	7.705
	(Age2)/100	The actual age of the head of household/year squared	23.522	7.226
Mediator variables
Human Capital	Education	No school = 1, primary school = 2, junior high school = 3, high school and technical secondary school = 4, college and above = 5	3.198	1.119
	Health status	Very Poor = 1, Relatively Poor = 2, Fair = 3, Relatively Good = 4, Very Good = 5	3.948	1.121
	Mulch film recycling capacity		2.960	1.109
Ecology Cognition	Perception of the severity of ground pollution	Very Low = 1, Relatively Low = 2, Fair = 3, Relatively High = 4, Very High = 5	4.261	0.900
	Recycling mulch film improves environmental awareness		4.392	0.873
	Awareness of recycled mulch film to increase production		4.334	0.942
	Awareness of the importance of plastic film pollution control to farmers		4.332	0.848
Moderator variable
Social norms	The degree of support of the family for the recycling of mulch film	Very Low = 1, Relatively Low = 2, Fair = 3, Relatively High = 4, Very High = 5	4.192	1.053
	The degree of support of neighbors for the recycling of plastic film		3.965	1.119
	The degree of support of village cadres for the recycling of plastic film		4.364	0.934
Control Variables
Party members and cadres	Whether there are party members or cadres in the family (No = 0, Yes = 1)		0.330	0.469
Part-time	Whether the domestic workforce is part-time (No = 0, Yes = 1)		0.600	0.490
Number of family members	Number of Persons in Household		3.610	1.405
The degree of salinization of cultivated land	There is no salinization = 1, there is a little salinization = 2, the degree of salinization is average = 3, the degree of salinization is large = 4, and the total salinization = 5		3.158	1.124
cooperative	Membership in a cooperative (no = 0, yes = 1)		0.648	0.470

. The questionnaire demonstrated good internal consistency, with a Cronbach’s Alpha coefficient of 0.860. The Kaiser–Meyer–Olkin (KMO) value was 0.721, indicating that the data were suitable for factor analysis. Principal component analysis with varimax rotation extracted a single common factor—labeled as the Social norms Factor—from the three observed variables. All item loadings exceeded 0.6. Furthermore, the composite reliability (CR) was greater than 0.8, and the average variance extracted (AVE) was above 0.5, providing evidence of satisfactory convergent validity for the measurement model.

(4) Control Variables: “Control variables include whether the farmer has a secondary occupation, the number of family members, the degree of soil salinization, and whether the farmer has joined a cooperative.

4. Results and Analysis

4.1. Mediation Effect Test

The regression results are shown in

Table 3. Hypothesis test results.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	Recycling Behavior	Recycling Behavior	Human Capital	Ecological Cognition	Human Capital	Ecological Cognition
Age	−0.000 (0.011)	0.232 ** (0.095)	0.128 *** (0.041)	0.106 *** (0.037)	0.127 *** (0.041)	0.091 ** (0.036)
Age squared term		−0.248 ** (0.101)	−0.130 *** (0.043)	−0.113 *** (0.039)	−0.129 *** (0.043)	−0.096 ** (0.038)
Social norms					0.046 *** (0.035)	0.147 *** (0.031)
An interaction item					0.012 (0.035)	0.077 *** (0.029)
Secondary interaction items					0.026 (0.036)	0.055 * (0.031)
Party members and cadres	−0.096 (0.169)	−0.090 (0.169)	−0.069 (0.073)	0.014 (0.066)	−0.069 (0.074)	−0.013 (0.065)
Part-time	−0.213 (0.170)	−0.172 (0.171)	0.077 (0.075)	−0.094 (0.067)	0.069 (0.075)	−0.108 (0.066)
Number of family members	0.103 (0.082)	0.097 (0.083)	0.004 (0.036)	0.223 *** (0.032)	−0.005 (0.036)	0.224 *** (0.032)
The degree of salinization of cultivated land	0.413 *** (0.085)	0.413 *** (0.086)	0.135 *** (0.037)	0.456 *** (0.033)	0.134 *** (0.037)	0.440 *** (0.033)
cooperative	0.169 (0.185)	0.171 (0.186)	0.163 ** (0.081)	0.118 (0.073)	0.143 * (0.082)	0.085 (0.071)
Constant terms	0.096 (0.580)	−5.211 ** (2.244)	0.969 (0.958)	2.455 *** (0.863)	1.001 (0.964)	−2.092 *** (0.854)
R2	0.074	0.357	0.218	0.353	0.051	0.385

Note: (1) Continuous variables were standardized using z-scores, whereas binary variables were retained in their original 0–1 form without standardization, the same below; (2) ***, **, * indicate significance at the 1%, 5% and 10% levels, respectively.

. In Model 1, the negative impact of age on farmers’ plastic film recycling behavior is not significant. However, in Model 2, when both age and age squared are included in the regression equation, age has a positive impact on farmers’ plastic film recycling behavior, while the coefficient for age squared is negative. Both are statistically significant, indicating an inverted U-shaped relationship between age and farmers’ plastic film recycling behavior, supporting Hypothesis 7. Similarly, the regression results in Model 3 show that there is also an inverted U-shaped relationship between age and human capital, supporting Hypothesis 2.

Combining the results from Models 2 and 3, it is evident that human capital mediates the impact of age on farmers’ plastic film recycling behavior, supporting Hypothesis 1. This implies that although young farmers have advantages in physical fitness and education level, their opportunity cost of farming is higher, leading them to invest more time in non-agricultural employment [33]. They are more inclined to choose extensive farming methods in agricultural production and management, lacking the motivation to engage in long-term investment activities such as plastic film recycling and improving farmland ecological environments. As farmers age, the “accumulation effect” of work experience in middle-aged farmers outweighs the “aging effect” of physical decline, enhancing their labor supply capacity, and leading to a relatively higher human capital stock compared to other age groups [9]. However, many companies impose age limits on hiring migrant workers, reducing non-agricultural employment opportunities for middle-aged farmers, making them more inclined to focus on farming and actively participate in plastic film recycling. On the other hand, although older farmers have improved work experience, their declining health and physical strength, along with limitations in their educational background, result in insufficient input of effective agricultural labor [34], leading to a reduced likelihood of participating in plastic film recycling.

In Model 4, age has a significant positive impact on ecological cognition, while age squared has a negative impact on ecological cognition, indicating an inverted U-shaped relationship between age and ecological cognition, supporting Hypothesis 4. Furthermore, combining the results from Models 2 and 4, it is clear that ecological cognition mediates the impact of age on farmers’ plastic film recycling behavior, supporting Hypothesis 3. Considering the current situation in rural areas, young farmers invest more energy in non-agricultural employment, have a low dependency on agriculture, and have a diminished sense of responsibility for protecting arable land. Their perception of the importance of managing plastic film pollution is weaker, and they are less motivated to participate in pollution control efforts. Middle-aged individuals, with more experience, benefit from the “learning-by-doing” spillover effect [35], making them more aware of the importance of plastic film recycling and more willing to engage in environmentally friendly agricultural decisions. For older farmers, whose children are grown, the “raising children for old age” period has shifted to a time of “reciprocal support”, where they place greater emphasis on enjoying family life and less on ecological benefits, leading to a reduced likelihood of participating in plastic film recycling [14].

4.2. Moderation Effect Test

According to Model 5, the interaction terms between age and social norms, as well as between age squared and social norms, do not have a significant positive effect on human capital, so Hypothesis 5 is not supported. However, Model 6 shows that the interaction terms between age and social norms, as well as between age squared and social norms, have a significant positive effect on ecological cognition, supporting Hypothesis 6.

Referring to the method proposed by [36] for testing the moderation effects in U-shaped relationships, this study establishes Model Equation (4) to examine the moderating effect of social norms on the U-shaped relationship between age and ecological cognition from three aspects: shape, inflection point, and overall level.

First, the impact of the moderating variable, social norms, on the curvature shape is examined. By taking the second derivative of $Age$ in Equation (4), the curvature K at the vertex of the quadratic function can be calculated (Equation (5)).

$$\mathrm{K}=2{\beta }_2+2{\beta }_{5}SN$$

(5)

For an inverted U-shaped curve, when K < 0, the smaller the value of K, the steeper the curve; conversely, the larger the value of K, the flatter the curve. The impact of the moderating variable, social norms, on the shape of the U-shaped curve can be assessed by taking the partial derivative of K with respect to social norms (SN), as shown in Equation (6)

$${\displaystyle \frac{\partial \mathrm{K}}{\partial \mathrm{S}\mathrm{N}}}=2{\beta }_5$$

(6)

According to the regression results from Model 6 in Table 1, β₅ = 0.055 is significant at the 10% level, indicating that the higher the level of social norms, the flatter the curve relationship between age and ecological cognition.

Second, the impact of the moderating variable, social norms, on the inflection point of the curve is considered. By taking the first derivative of ${Age}^{*}$ in Equation (5) and setting it equal to 0, the inflection point can be calculated as:

$${Age}^{*}=-{\displaystyle \frac{{\beta }_1+{\beta }_{4}SN}{2({\beta }_2+{\beta }_{5}SN)}}$$

(7)

Taking the partial derivative of ${Age}^{*}$ with respect to SN.

$${\displaystyle \frac{\partial {Age}^{*}}{\partial SN}}={\displaystyle \frac{{\beta }_1{\beta }_5-{\beta }_2{\beta }_4}{{2({\beta }_2-{\beta }_{5}SN)}^2}}$$

(8)

The inflection point of the curve, denoted as ${Age}^{*}$, can be obtained from Equation (7). If the partial derivative of ${Age}^{*}$ with respect to the moderating variable SN is greater than 0, then as SN increases, ${Age}^{*}$ increases, causing the inflection point of the curve to shift to the right. Conversely, if the partial derivative is less than 0, then as SN increases, ${Age}^{*}$ decreases, causing the inflection point to shift to the left. According to Equation (8), since the denominator is greater than 0, the impact of the moderating variable SN on the axis of symmetry of the curve is determined by ${\beta }_1{\beta }_5-{\beta }_2{\beta }_4$. In Model 6 of Table 1, β₁, β₂, β₄, β₅ are all significant. Substituting the coefficients into the calculation gives ${\beta }_1{\beta }_5-{\beta }_2{\beta }_4>0$, indicating that as the moderating variable increases, the axis of symmetry of the curve shifts to the right.

Third, the impact of the moderating variable on the overall level of the curve is considered. The value of the dependent variable at any given value of the independent variable depicts the change in the overall level of the curve. Let the ecological cognition under high social norms be denoted as ECO_SNH, and the ecological cognition under low social norms be denoted as ECO_SNH. Then:

$$\begin{array}{l}{ECO}_{SNH}-{ECO}_{SNL}=\left({\beta }_0+{\beta }_{1}Age+{\beta }_2{\left(Age\right)}^2+{\beta }_3{SN}_H+{\beta }_4{SN}_H\times Age+{\beta }_5{SN}_H\ast {\left(Age\right)}^2+X+{\mu }_i\right)\\ -{(\beta }_0+{\beta }_{1}Age+{\beta }_2{\left(Age\right)}^2+{\beta }_3{SN}_L+{\beta }_4{SN}_L\times Age+{\beta }_5{SN}_L\ast {\left(Age\right)}^2+X+{\mu }_i)\\ =({SN}_H-{SN}_L)({\beta }_5{\left(Age\right)}^2+{\beta }_{4}Age+{\beta }_3\end{array}$$

(9)

Based on Equation (9), since ${SN}_H-{SN}_L$, the difference ${ECO}_{SNH}-{ECO}_{SNL}$ is determined by ${ECO}_{SNH}-{ECO}_{SNL}$. Let the quadratic function $f\left(Age\right)={\beta }_5{\left(Age\right)}^2+{\beta }_{4}Age+{\beta }_3$. From Model 7 in Table 1, we know that β5 = 0.055, which is significant at the 10% level, and the parabola opens upwards. Both β4 and β3 are significant at the 1% level, and $\nabla ={{\beta }_4}^2-4{\beta }_5{\beta }_3=-0.026<0$. Therefore, regardless of the value of SN, ${SN}_H-{SN}_L$ is always greater than 0, indicating that a high level of social norms raises the overall level of the inverted U-shaped relationship between age and ecological cognition.

In summary, the higher the level of social norms, the flatter the inverted U-shaped curve between age and ecological cognition, with the inflection point shifting to the right. This elevates the overall level of the curve relationship, indicating that social norms positively moderate the relationship between age and ecological cognition, thus supporting Hypothesis 6. Furthermore, as shown in Figure 2, compared to the inverted U-shaped curve at SN − M, the curve at SN + M exhibits a slight shift in the inflection point and becomes steeper, indicating that social norms amplify the intensity of age’s impact on ecological cognition., thus supporting Hypothesis 6.

4.3. Heterogeneity Test

The impact of agricultural labor age on farmers’ plastic film recycling behavior may be related to the quality of arable land and whether they have joined a cooperative. (1) In this study, the degree of soil salinization is used as a proxy variable for arable land quality. The samples were divided into two groups, representing high and low land quality, with the average degree of salinization of 3.158 as the boundary. Regression analysis is conducted for each group, and the results are shown in Models 7 and 8 of

Table 4. Heterogeneity test.

Variables	Model 7	Model 8	Model 9	Model 10
	High Quality of Cultivated Land	Low Arable Land Quality	Join a Cooperative	Not a Member of a Cooperative
Age	0.272 ** (0.136)	0.150 (0.141)	0.224 ** (0.108)	0.248 (0.200)
Age squared term	−0.273 * (0.144)	−0.179 (0.149)	−0.242 ** (0.115)	−0.257 0.213
Fisher test experience p value	0.000 ***		0.000 ***
Control Variables	Controlled	Controlled	Controlled	Controlled
Constant terms	−6.549 ** (3.235)	−2.687 (3.326)	−4.793 * (2.550)	−5.764 (4.712)
Number of samples	375	364	556	183
R2	0.212	0.082	0.235	0.264

Note: (1) ‘Empirical p value’ is used to test the significance of the difference in coefficients between groups, which is calculated by Fisher combination test (sampling 2000 times). (2) ***, **, * indicate significance at the 1%, 5% and 10% levels, respectively.

. The age of agricultural labor has a significant impact on farmers’ plastic film recycling behavior in both groups. However, the inflection point for high-quality land is 49.82, while for low-quality land it is 41.19, indicating that the inflection point for agricultural labor age is delayed in higher-quality land compared to lower-quality land. A possible reason is that better land conditions may offset the loss of agricultural labor due to aging, which also indirectly underscores the importance of implementing high-standard farmland policies across the board. (2) When grouping the sample based on whether the farmers have joined a cooperative, the results are shown in Models 9 and 10 of Table 4. For farmers who have not joined a cooperative, the age of agricultural labor does not significantly affect their plastic film recycling behavior. This could be because, for non-cooperative farmers, government economic subsidies and policy promotion, rather than age or physical strength, are more critical factors in their decision-making regarding plastic film recycling. In this study, Fisher ‘s test was used to analyze the differences between groups. In order to verify the reliability of the group regression results, Fisher test was used to analyze the differences between groups. The results showed that the empirical p value of the Fisher test was less than 0.1, indicating that there was a significant difference in the coefficient between the groups.

4.4. Robustness Test

(1)

Grouped Regression: according to “inverted U-shaped” relationship diagram, the left side represents a positive effect, and the right side represents a negative effect. Consequently, the sample is divided into left and right groups centered around the inflection point. The results for the left-side sample can be seen in Models 11 and 12 of

Table 5. Robustness test.

Variables	Model 11	Model 12	Model 13	Model 14	Model 15
	Age < 51.48	Age ≥ 51.48	OLS	Tail Reduction	Probit
Age	0.032 ** (0.016)	−0.095 * (0.054)	0.054 ** (0.022)	0.243 ** (0.097)	0.238 ** (0.095)
Age squared divided by 100			−0.058 ** (0.023)	−0.258 ** (0.103)	−0.255 ** (0.101)
Party members and cadres	−0.140 (0.194)	−0.022 (0.340)	−0.021 (0.040)	−0.093 (0.169)	0.074 (0.170)
Part-time	−0.137 (0.202)	−0.059 (0.349)	−0.040 (0.040)	−0.168 (0.172)	0.160 (0.172)
Number of family members	0.090 (0.092)	0.095 (0.192)	0.022 (0.019)	0.097 (0.083)	0.094 (0.083)
The degree of salinization of cultivated land	0.419 *** (0.097)	0.457 ** (0.196)	0.098 *** (0.020)	0.414 *** (0.086)	0.359 *** (0.079)
cooperative	0.156 (0.212)	0.205 (0.391)	0.040 (0.044)	0.170 (0.186)	0.147 ** (0.066)
Constant terms	−1.256 (0.792)	5.249 * (3.139)	−0.712 (0.519)	−5.469 ** (2.297)	−7.363 *** (2.260)
Number of samples	570	169	739	739
R2	0.261	0.236	0.253	0.264	0.262

Note: ***, **, * indicate significance at the 1%, 5% and 10% levels, respectively.

. When age is less than 51.48 years, the coefficient for age is significantly positive at the 5% statistical level, proving that the age of agricultural labor positively in fluences farmers’ plastic film recycling behavior in the left-side sample. Similarly, as shown in Model 12 for the right-side sample, when age is greater than or equal to 51.48 years, the coefficient for age is significantly negative at the 1% statistical level, indicating that the age of agricultural labor has a negatively impact on farmers’ plastic film recycling behavior in the right-side sample. In summary, the coefficients on both sides of the inflection point are opposite, and their absolute values are very close, further confirming the robustness of the inverted U-shaped relationship between age and farmers’ plastic film recycling behavior.

(2)

Changing the Econometric Model: Given the binary nature of the dependent variable, the model is re-estimated using an OLS model. As shown in Model 13 of Table 5, the OLS model results show a positive coefficient for age and a negative coefficient for age squared, both of which are statistically significant. Similarly, in Model 15 where the probit model was adopted, the age coefficient is positive while the squared age term shows a negative coefficient, both of which are statistically significant. This confirms the “inverted U-shaped” relationship and further validates the stability of the results.

(3)

Truncation Treatment: There may be outliers in the sample that could affect the empirical results. Therefore, a 1% truncation is applied to the lower and upper ends of the age distribution. The results, shown in Model 14 of Table 5, are consistent with those before truncation, once again verifying the robustness of the results.

5. Discussion

5.1. Overview and Comparison

Based on the empirical analysis of survey data from 739 farmers in Xinjiang, this study reveals an inverted U-shaped relationship between the age of the agricultural labor force and farmers’ plastic film recycling behavior, which aligns with the associations reported by Ullah et al. [4] regarding technology adoption among cotton farmers in Pakistan. Conversely, this result differs from Liu et al. [37], who observed a linear negative association between age and technology adoption in rice-producing regions of China. This difference may reflect crop-specific labor intensity (plastic-mulched cultivation demands higher physical effort) and regional labor structure heterogeneity (Xinjiang exhibits a higher proportion of large-scale farming operations). Figure 1 illustrates the dual-pathway mechanism through which age is associated with farmers’ recycling behavior, primarily operating through the human capital pathway and the ecological cognition pathway. These two mechanisms jointly shape farmers’ recycling participation at different stages of their life cycle, explaining the observed inverted U-shaped pattern between age and recycling behavior.

Human Capital Pathway. During early and middle adulthood, farmers typically exhibit higher levels of technical learning capacity, adaptive ability, and information access, which significantly enhance their ability to engage in green agricultural practices. At this stage, farmers are more actively involved in production activities, enabling them to effectively integrate recycling technologies into their farming systems. Empirical evidence confirms that accumulated experiential knowledge and production skills substantially increase the likelihood of adopting environmentally friendly technologies [38]. When age exceeds a certain range, cognitive flexibility begins to decline [39], and access to information also becomes more constrained (e.g., smartphone usage declines significantly), which collectively weakens their ability to adopt and sustain recycling practices. This life-cycle pattern provides a micro-level explanation for why participation in recycling behavior peaks during middle age. Importantly, this mechanism aligns well with human capital theory: middle-aged farmers (approximately 35–55 years old) are more likely to invest in green technologies due to their intensive engagement in agricultural production [40] and strong sense of intergenerational responsibility [41]. These findings highlight how accumulated knowledge, social obligation, and adoption propensity interact to reinforce behavioral engagement.

Ecological Cognition Pathway. Age is also closely related to farmers’ perceptions of environmental risk and ecological responsibility. Older farmers generally demonstrate higher levels of environmental awareness and stronger perceptions of ecological risk [42], which theoretically should encourage greater recycling participation. However, this elevated ecological cognition does not fully translate into actual behavioral engagement. Physical constraints (e.g., reduced daily labor capacity), shorter expected payback horizons, and stronger present-bias preferences limit older farmers’ ability to act on their environmental concerns. This “cognition–behavior gap” is consistent with the behavioral economics perspective, which suggests that environmental awareness alone is insufficient to drive behavioral change when individuals face capability and temporal constraints [43]. Taken together, these two pathways reveal the dynamic interaction between capability and motivation across the life course. Middle-aged farmers possess both relatively strong technical capability and moderate-to-high ecological awareness, resulting in the highest recycling participation in this group. In contrast, younger farmers often lack sufficient human capital, while older farmers—despite their heightened environmental awareness—face declining physical capacity and reduced incentive horizons, leading to lower behavioral conversion rates.

Revisiting the Moderating Role of Social Norms. Social norms exhibited a significant positive moderating association with the relationship between farmers’ age and ecological cognition, supporting Ostrom’s [44]. collective action framework and aligning with findings from Nepal [45]. This indicates that social norms can enhance the translation of ecological awareness into actual recycling behavior by shaping group expectations and behavioral standards. However, social norms did not exhibit a significant moderating association with the relationship between farmers’ age and human capital, in contrast to Tang et al. [10] in North China. This difference may be attributed to the relatively lower density of formal agricultural organizations and social networks in the study region, which limits the channels through which social norms can amplify human capital effects. In the absence of efficient information-sharing platforms and collaborative structures, even farmers with substantial experience and skills may not fully benefit from the reinforcing role of social norms. These findings suggest that relying solely on social norms to promote green technology adoption may be insufficient. Complementary measures, such as strengthening agricultural cooperatives and expanding information-sharing networks, are necessary to provide institutional and social support for the effective utilization of human capital.

5.2. Limitations and Future Directions

First, the data used in this study are cross-sectional, which limits our ability to establish causal associations between labor age, human capital, ecological cognition, and mulch film recycling behavior. Future longitudinal or experimental studies would be valuable to provide stronger causal evidence. Second, while this study examines the moderating role of social norms, it does not fully capture the multi-dimensional nature of social capital and network structures. Subsequent studies could incorporate more detailed measurements of social networks, trust, and cooperative behavior to better understand how collective mechanisms influence technology adoption. Third, the findings of this study specifically reflect the recycling behavior of farmers in Xinjiang, and the results may not be directly applicable to all farmers in China, as agricultural practices, labor structures and ecological cognition vary significantly across regions. Future research should examine these relationships in other regions to better understand the generalizability of these findings. Moreover, the study uses a binary dependent variable (recycling behavior: yes/no), which simplifies the nature of the outcome. Future research may explore more nuanced measures of recycling behavior, such as the intensity or frequency of participation, to provide deeper insights into the dynamics of adoption. The findings of this study suggest that while age, human capital, and ecological cognition are important drivers of recycling behavior, practical barriers such as the availability of recycling facilities, transportation, and associated costs also need to be addressed. Future research should examine how these factors influence recycling participation and explore strategies to reduce barriers. Finally, with the rapid development of digital agriculture and smart farming technologies, future studies should explore how digital transformation reshapes farmers’ behavioral patterns, particularly for different age cohorts. Such research would offer valuable insights for designing adaptive and inclusive environmental policies.

6. Conclusions and Implications

(1)

The Relationship Between Labor Age and Plastic Film Recycling Behavior Exhibits an Inverted U-Shape: This indicates that middle-aged farmers have become the primary force in participating in green agricultural production practices. The government should promote plastic film recycling behavior among different types of farmers in a targeted manner. For young farmers, efforts should be made to accelerate the training of professional farmers, enhance the appeal of farming as a career, and attract more young people to specialize in agricultural production. For middle-aged farmers, the focus should be on providing support in terms of technology and funding to encourage their participation in plastic film recycling and to leverage their role as examples in this behavior. For elderly farmers, it is essential to strengthen production and living subsidies to incentivize their participation in plastic film recycling.

(2)

Human Capital and Ecological Cognition Are Important Mediating Variables. These factors mediate the influence of age on farmers’ behavior in recycling plastic film. Both human capital and ecological cognition demonstrate a pattern of initially increasing and then decreasing with age. Therefore, it is necessary to further enhance the social security system to improve the quality of human capital within the agricultural labor force. This will create favorable conditions for their production and management activities. Moreover, greater efforts should be made to promote policies for plastic film recycling, utilizing internalization and value orientation to enhance farmers’ ecological cognition and thus encourage their active participation in plastic film recycling.

(3)

Social norms moderate the relationship between age and ecological cognition: as the level of social norms increases, the inverted U-shaped curve between age and ecological cognition becomes more gradual. The inflection point of the curve shifts to the right, resulting in an overall higher level of the relationship. However, social norms do not significantly moderate the relationship between age and human capital. Therefore, to promote plastic film recycling, it is important to leverage the network effects created by rural social norms. This includes actively cultivating and expanding social capital for individual laborers, promoting knowledge dissemination, enhancing individual ecological cognition, and increasing the likelihood of farmers participating in plastic film recycling.