Training and Self-Learning: How to Improve Farmers’ Willingness to Adopt Farmland Conservation Technology? Evidence from Jiangsu Province of China

Abstract

The quality of cultivated land is an important basis for guaranteeing grain yield, and improving the quality of cultivated land is an important initiative of the Chinese government to implement its food security strategy. This paper explores the effects of formal technical training and informal online self-directed learning on farmers’ willingness to adopt cultivated land conservation technology using farmer-level survey data. The results show that farmers’ participation in technical training effectively increased their willingness to adopt straw return technologies, but participation in online self-directed learning did not affect farmers’ willingness to adopt, while farmers who participated in both technical training and online self-directed learning had stronger willingness to adopt. These results show that internet-based self-directed learning is a useful supplement to the formal training system. Further analysis revealed that participation in technical training increased farmers’ awareness of the usefulness of the straw-return technology, which in turn increased their willingness to adopt the conservation technology of cropland. This paper clarifies the impact and mechanisms of the two main existing technology learning modes on farmers’ adoption of new technologies in rural China and provides a reference for the promotion of agricultural technology diffusion and the improvement of the quality of cultivated land in China.

1. Introduction

Cultivated land is the foundation of agricultural production and an important material guarantee to promote the sustainable development of agriculture. Under the condition of an increasing population and less land, China’s agricultural production has been adhering to the model of high input and high yield. The long-term intensive use of cultivated land leads to a low content of soil organic matter and a decline in basic fertility. In 2019, China’s Ministry of Agriculture and Rural Affairs conducted a nationwide survey on the quality of cultivated land. In this survey, the country’s arable land was divided into ten grades in descending order of quality, with the first three grades being high-quality arable land. The results of this survey showed that the average grade of cultivated land was 4.76, and only 31.24% of the country’s cultivated land was of high quality. Good quality of cultivated land is conducive to increasing grain yield and achieving high-quality agricultural development. Therefore, strengthening quality construction and protection of cultivated land is of positive significance for ensuring China’s food security.

Straw is an important by-product of crop production, and in China, farmers have a historical tradition of burning straw [1], which they believe to be beneficial in reducing pests and diseases on the land and, more importantly, an economically inexpensive way to dispose of straw [2,3] that can be put back into the next crop more quickly. However, the fact is that long-term straw burning is a very low-cost way to dispose of straw. However, long-term straw burning can lead to land impoverishment [4] and straw burning has a negative impact on regional air quality [1,5,6]. In order to solve the above problems, the Chinese government has been promoting straw-return technology since 2008. Straw-return technology has the following two advantages: on the one hand, it is beneficial to the improvement of land quality, and on the other, is beneficial to environmental protection [7]. In reality, farmers are the main body of land conservation, and the question of how to motivate farmers to adopt arable land conservation technology is an important issue of concern for the government.

There has been a rich discussion in the academic community around the above-mentioned topics. Most of the existing studies take straw-return technology as an example and analyze it mostly from the perspective of farmers’ individual endowment characteristics and farming operation characteristics, exploring socialization services [8], time preference [9], the stability of land management rights [10], scale of farmland [11], land transfer [12], land fragmentation [13], and age and education level [14,15] on the adoption of conservation technologies by farmers.

At the policy level, in order to incentivize farmers to adopt straw-returning technology, the Chinese government provides subsidies for farmers to adopt straw returning to reduce their technology adoption costs. It has also implemented a relatively strict no-burning policy to monitor farmers’ straw burning behavior. Although the implementation of these two policies has been effective in promoting the adoption of straw-returning technology by farmers [16,17], the Chinese government faces a high financial burden in promoting the use of the technology by farmers. The government expects farmers to consistently adopt straw-returning technology, but the administrative push from the government alone is not enough and is not economically efficient. Theories and studies have shown that farmers’ awareness of agricultural technologies is a key factor influencing their adoption [12,18,19], and when farmers’ awareness of new technologies increases, it will help increase their endogenous motivation to adopt new technologies.

In reality, the Chinese government often provides technical training to farmers when promoting new technologies, with a view to improving their understanding of the new technologies. At the same time, with the application of smartphones by farmers, the village also provides an informal knowledge sharing channel, where its agricultural technology station will invite some farmers to join a technology sharing group on WeChat, and where agricultural technology service staff will share knowledge related to straw returning in the chat group from time to time. However, the model mainly relies on farmers for independent learning. Compared with formal technical training, the cost of using information technology to disseminate new agricultural technology will be much lower. At the same time, farmers can learn new technologies in their leisure time, and the training format is more flexible. Therefore, do technical training and informal online self-learning help increase farmers’ willingness to adopt straw returning technologies? Are the two complementary or alternative relationships? What are the mechanisms that influence farmers’ willingness to adopt? This paper will answer these questions, and the findings will provide suggestions for improving the diffusion of conservation technologies and for improving the current policies supporting conservation technologies.

Compared with the existing studies, this paper may have some contributions in the following two aspects: (1) Crop straw resource utilization is not sufficiently developed in rural areas of China, and in the context of a strong government policy against burning and the absence of straw resource utilization by farmers, farmers have no choice but to return straw to the fields. This means that it is difficult to identify farmers’ real thoughts on the adoption of this technology by their behavior of whether or not they return straw to the field, and it is also difficult to clarify other influencing factors that affect farmers’ willingness to return straw to the field. Therefore, this paper will explore the key factors influencing farmers’ adoption of straw-returning technology by examining the farmers’ willingness to adopt it. (2) This paper incorporates technical training provided by government departments and informal web-based self-directed learning extension into a unified analytical framework and identifies the effects and mechanisms of both on farmers’ willingness to return straw to the field.

The rest of the paper is organized as follows: Section 2 provides hypothesis development. Section 3 describes our data and provides econometric specification. Section 4 presents our empirical results. Firstly, we estimate the impact of farmers’ participation in formal technical training and informal online self-directed learning on farmers’ willingness to return straw to the field, then we discuss the underlying mechanisms of the technical training and informal self-learning. Finally, the heterogeneity of the effects is estimated across famer characteristics. Section 5 concludes the paper and provides policy implications.

2. Theoretical Analysis and Research Hypothesis

It is assumed that farmers face two agricultural production modes, one in which they choose to return straw to the field in agricultural production and one in which they do not choose to return straw to the field. Under the two different agricultural production modes, since the straw returned to the field will have a positive impact on the quality of the cultivated land, and may lead to differences in the amount of production factors input by the farmers, the income function of the farmers under the two different agricultural production modes is written in the following form:

$${\pi }_s=P_s\times Q_s-{{\displaystyle \sum }}_{n=1}^m{W}_s\times X_s$$

(1)

$${\pi }_t=P_t\times Q_t-{{\displaystyle \sum }}_{n=1}^m{W}_t\times X_t$$

(2)

Equation (1) is the agricultural business income ${\pi }_s$ if farmers adopt the straw-returning technology in agricultural production, and Equation (2) is the agricultural business income ${\pi }_t$ if farmers do not adopt the straw-returning technology in agricultural production. $P_s$ and $Q_s$ are the prices and yields of agricultural products after farmers adopt the straw-returning technology, and $W_s$ and $X_s$ are the quantities and prices of various agricultural production factors. $P_t$ and $Q_t$ are the prices and yields of agricultural products if farmers do not adopt the straw-returning technology, and $W_t$ and $X_t$ are the quantities and prices of various agricultural production factors when farmers do not adopt the straw-return technology. The willingness of farmers to adopt straw-return technology depends on the expected business returns under the two technology models.

$${\pi }^{\prime }={\pi }_s-{\pi }_t=(P_s\times Q_s-{{\displaystyle \sum }}_{n=1}^m{W}_s\times X_s)-(P_t\times Q_t-{{\displaystyle \sum }}_{n=1}^m{W}_t\times X_t)$$

(3)

We further assume that π’ is the difference in expected benefits between the scenarios where farmers adopt and do not adopt straw-returning technology, as shown in Equation (3). When the difference in expected benefits is greater than 0, then farmers will choose to adopt the technology, while when the benefits from adopting the technology are greater than the costs, then farmers will not choose to adopt the technology. When the difference in expected benefits is larger, the stronger the willingness of farmers to adopt the straw-return technology.

$$Adop{t}_i=\left\{\begin{array}{c}0, {\pi }^{\prime }<0\\ 1, {\pi }^{\prime }\ge 0\end{array}\right.$$

(4)

When farmers are unaware of the new technology, they may not realize the potential positive impact of adopting straw-return technology on the quality of their farmland. When farmers attend training on straw-returning technology, professional agricultural technicians will explain the advantages of returning straw to the field, which means that the adoption of straw-returning technology will improve the quality of farmland and reduce the input of fertilizers and other factors of production [20,21], while the improvement of farmland strength will also improve crop yields, and systematic technical training will help improve farmers’ awareness of the benefits of straw-returning technology. The higher the level of farmers’ awareness of the benefits of the technology, the higher the expected benefits of adopting the technology, and the higher the willingness of farmers to adopt the technology. When farmers realize that returning straw to the fields will improve the expected benefits of adopting the technology due to the improvement of the quality of the cultivated land, they are more willing to adopt the straw-returning technology. According to Equation (3), when farmers expect to increase their expected returns from adopting the straw-return technology, the likelihood of ${\pi }^{\prime }$ > 0 will be higher and their likelihood of adopting the straw-return technology will increase accordingly. Davis (1989) proposed a technology acceptance model based on rational behavior theory, which is widely used in the field of technology adoption [18]. In the technology acceptance model, users’ willingness to adopt a new technology is influenced by the perceived usefulness of the technology. In the existing studies, the ability to improve users’ performance or income is an important indicator of the perceived usefulness of technology [22]. Therefore, we argue that when farmers participate in technology training, it increases their level of the perceived usefulness of the straw-returning technology and further increases their willingness to adopt the technology.

With the gradual penetration of ICT in rural areas of China, information technology tools such as computers and cell phones have gradually become popular in rural areas. It has been found that ICT helps to reduce the role of information disadvantage of farmers and gradually becomes an important means for farmers to obtain information [23,24,25]. In the technology sharing group, farmers mainly read articles and videos related to straw-returning technology to increase their understanding of the new technology, which can improve their knowledge of the new technology to a certain extent. In terms of different ways of learning new technologies, farmers’ cognitive understanding of the usefulness of new technologies after participating in technical training may be better than who were self-taught for the following reasons: first, farmers need to study the technical articles in the technology sharing group independently after joining, and the effect of self-learning new technologies is closely related to the level of personal learning ability, which puts higher demands on farmers’ learning ability. In reality, the education level of Chinese farmers is not high, and most of them have a junior high school education [26], so that self-learning may not be a good way for farmers to learn new technologies. Secondly, when farmers participate in technical training, there are special agricultural technicians to introduce the straw-returning technology systematically and also to answer farmers’ questions related to the adoption of the new technology. However, in the technical sharing group, farmers have difficulty in getting professional answers, which may have an impact on enhancing farmers’ perception of the usefulness of the new technology. At the same time, the straw-return technology requires farmers to pay the corresponding technology adoption costs, as, though government subsidies exist, they are not sufficient to cover the full cost of technology adoption and farmers still need to pay the additional amount. It is difficult to reduce farmers’ risk perception of the new technology when they do not have sufficient knowledge of it, and it is also difficult to improve farmers’ expected business returns after adopting straw returning technology, which means it is hard to improve the ${\pi }^{\prime }$ in (3), and is in turn not conducive to improving farmers’ willingness to adopt straw-returning technology.

Because of the above reasons, we believe that farmers’ participation in on-site technical training will have a better effect than self-learning, while the effect of joining the WeChat technical sharing group on improving farmers’ technical knowledge may be limited or even absent. However, for farmers who have participated in the training, they already have a certain cognitive understanding of straw-returning technology, and participating in the technology sharing group will help further strengthen their understanding of straw-returning technology and their cognition of the technology will be more comprehensive. This can be a useful supplement to the formal training, and therefore be more likely to enhance farmers’ willingness to adopt straw regranting technology.

Based on the above analysis, we propose the following hypotheses:

Hypothesis 1 (H1).

Participation in straw-returning technology training helps to improve farmers’ cognitive understanding of the technology, and further promote farmers’ willingness to adopt the it.

Hypothesis 2 (H2).

The effect of participation in online self-directed learning on farmers’ willingness to adopt is uncertain, but for farmers who have already participated in the training, continued participation in online self-directed learning is more beneficial in increasing farmers’ willingness to adopt.

3. Data, Variables, and Model

3.1. Data Sources

The data used in this paper are from a field survey conducted by the “Straw resource utilization” group of Nanjing Agricultural University in six counties in Jiangsu Province in 2016. The survey covered the northern, central and southern regions of Jiangsu Province, as shown in Figure 1. The group randomly selected three towns in each county, two villages in each town, and 10–12 people in each village, and obtained a total of 382 questionnaires. After eliminating the samples with missing relevant required information, 375 valid questionnaires were finally obtained.

The questionnaires included data on basic personal information, household structure characteristics, agricultural production input and output, straw-returning technology awareness and farmers’ willingness to adopt straw-returning technology. Jiangsu Province is located in the eastern region of China and is a strong economic province in China. Similarly, Jiangsu is at the forefront of China for agricultural modernization, and its experience may provide a reference for the whole country [27]. Therefore, the above data can provide better support for the empirical analysis of the subject of this paper.

3.2. Model Specification

The dependent variable in this paper is farmers’ willingness to return straw to the field. According to the basic view of ordinal utility theory, willingness to adopt is a psychological feeling, is difficult to measure directly and can only be measured by ranking. Considering the characteristics of the data, the dependent variable is ranked data, and in order to test the impact of technical training, on farmers’ willingness to adopt straw-returning technology, the ordered probit model was selected for empirical analysis. Therefore, the following econometric model was developed.

$$Strawretur{n}_i={\alpha }_0+{\alpha }_{1}trai{n}_i+{\alpha }_{2}wecha{t}_i+\beta X_i+{\epsilon }_i$$

(5)

$Strawretur{n}_i$ is a potential, unobservable and real subjective feeling of farmer i. In this paper, the five-point Likert method is used to characterize farmers’ willingness to adopt straw-returning technology, $Strawreturn=1,2,3,4,5$, which means “strongly opposed”, “somewhat opposed”, “no effect”, “somewhat agree” and “strongly agree” respectively, with higher values indicating stronger willingness of farmers to adopt. $r_i$ indicates the cut-off point of $Strawretur{n}^{\ast }$, if $Strawreturn{ }^{\ast }\le r_0$, then Strawreturn = 1; if $r_0<$ $Strawretur{n}^{\ast }\le r_1$, then Strawreturn = 2, and so on, when $r_3<Strawretur{n}^{\ast }\le r_4$, then Strawreturn = 5, and so on.

$$Strawreturn=\left\{\begin{array}{c} 1, if Strawretur{n}^{\ast }\le r_0\\ 2, if r_0<Strawretur{n}^{\ast }\le r_1\\ 3, if r_1<Strawretur{n}^{\ast }\le r_2\\ 4, if r_2<Strawretur{n}^{\ast }\le r_3\\ 5, if r_3<Strawretur{n}^{\ast }\le r_4\end{array}\right.$$

(6)

$Trai{n}_i$ refers to whether farmers participate in straw returning training, $Wecha{t}_i$ refers to whether farmers participate in online independent learning. $X_i$ is the set of other control variables affecting farmers’ straw return, and ${\epsilon }_i$ is a random disturbance term.

3.3. Variables and Descriptive Statistics

3.3.1. Dependent Variables

The dependent variable in this paper is farmers’ willingness to adopt straw returning technology, which is measured by the question “Are you willing to return straw to the field”, and farmers choose a value between 1 and 5.

3.3.2. Independent Variables

The independent variable in this paper is whether the farmers have participated in the technical training of straw field return, and the value is 1 if the farmers have participated, and 0 if they have not. Similarly, if the farmer participates in the WeChat group for self-learning straw-return technology, the value is assigned to 1, and vice versa is 0.

In terms of control variables, this paper refers to Sun et al., (2019), Lu et al., (2020) and Cheng et al., (2022) to control for farm household demographic characteristics [8,17,28], farm household characteristics, straw subsidy policy and straw burning ban policy to influence farm households’ willingness to return straw to the field. For the demographic characteristics of farm households, we selected three variables, namely, gender, age, and education level, to measure the individual characteristics of farm households. In terms of farm household characteristics, the number of household laborers, the proportion of laborers engaged in non-farm work, and the total planted area of the household were selected to capture the household-level characteristics of farm households. Farmers’ willingness to return straw to the fields is also influenced by relevant policies, so we also controlled for straw return subsidy and local government monitoring variables for straw burning, using whether the village chief paid a straw burning deposit to the higher level and whether cameras were installed in the fields, respectively.

The definitions of the variables involved in the empirical study and the descriptive statistics are shown in

Table 1. Descriptive statistics of variables.

Variables	Definition	Mean	S.D.
Straw returning	The willingness of the famer to implement straw retuning (1 = strongly opposed, 2 = somewhat opposed, 3 = no effect, 4 = somewhat agree, 5 = strongly agree)	3.18	1.60
Technical training	Whether farmers participate in straw returning training (1 = yes and 0 = no)	0.27	0.46
Self-learning	Whether farmers participate in the WeChat group chat of the village agricultural technology station (1 = yes and 0 = no)	0.15	0.35
Age	Age of farmers (year)	58.47	9.75
Gender	Gender of farmers (1 = male and 0 = female)	0.91	0.29
Education	Years of education for famers	6.26	3.72
Labor	Number of family labor	2.47	1.24
Farm size	Size of cultivated land (mu)	60.85	157.44
No. of plots	Number of plots planted by the household	3.66	2.67
Rented land	Share of rented in land (%)	0.36	0.59
Subsidy	The amount of subsidies for returning straw to the field (yuan per mu)	19.41	11.94
Camera monitor	Whether camera is installed to monitor fires (1 = yes and 0 = no)	0.33	1.03
Cash deposit	Whether village leaders were required to deposit cash to town government (1 = yes; 0 = no)	0.72	0.45
Off farm	The proportion of households with off-farm employment (%)	0.40	0.32

. The average willingness to adopt straw-returning technology of the sample farmers was 3.18, 27% of the farmers had participated in straw-returning technology training, and 9% of the farmers had participated in technology sharing groups on WeChat at agricultural technology stations.

According to the different types of farmers, we further divided the farmers into those who had participated in technology training and those who had not, and those who had joined the WeChat independent learning group and those who had not. By calculating the mean values of the willingness to adopt straw-returning technology among different types of farmers and conducting a t-test for the mean values. The results of the test are shown in

Table 2. Comparison of adoption willingness of different types of farmers.

	(1)	(2)	(3)	(4)	(5)	(6)
	Technical Training		The Differences	WeChat Group		The Differences
	Yes	No	t-Statistic	Yes	No	t-Statistic
Willingness	3.412	3.082	0.331 *	3.167	3.129	0.038

Note: * indicate statistical significance at 10%.

. The adoption intention of farmers who participated in the technology training was significantly higher than those who had not participated in the training, and the difference in means was significant at the 10% level. Meanwhile, the adoption intention of farmers who participated in the village agricultural technology station’s WeChat group was slightly higher than that of farmers who did not participate in the WeChat group, but there was no significant difference in the mean difference between them. Strictly speaking, we can only consider the adoption intention of farmers who participated in technology training to be significantly higher than that of farmers who did not participate in training if the t-test of the mean between the two groups is significant at the 5% level. Of course, the above analysis only serves to show that there may be some differences in farmers’ willingness to adopt different types of technology learning methods, but whether there is a causal relationship between the two groups requires a more rigorous econometric analysis.

4. Empirical Results

4.1. Baseline Result

In this paper, the willingness of farmers to adopt straw-returning technology was firstly considered as a continuous variable and firstly regressed using OLS model and the regression results are shown in columns (1) of

Table 3. Baseline regression results.

	(1)	(2)	(3)	(4)	(5)
Variables	OLS	Ordered Probit	Ordered-Probit-Margins
			1	3	5
Technical training	0.448 **	0.368 ***	−0.111 ***	−0.005 *	0.126 ***
	(0.185)	(0.141)	(0.042)	(0.003)	(0.047)
Self-learning	0.047	0.045	−0.014	−0.001	0.015
	(0.373)	(0.287)	(0.087)	(0.004)	(0.098)
Age	−0.001	−0.000	0.000	0.000	−0.000
	(0.010)	(0.007)	(0.002)	(0.000)	(0.002)
Education	0.020	0.012	−0.003	−0.000	0.004
	(0.025)	(0.017)	(0.005)	(0.000)	(0.006)
Gender	−0.274	−0.176	0.053	0.003	−0.060
	(0.297)	(0.199)	(0.060)	(0.003)	(0.068)
Labor	−0.134	−0.086	0.026	0.001	−0.030
	(0.084)	(0.058)	(0.018)	(0.001)	(0.020)
Farm size	−0.135 **	−0.082 *	0.025 *	0.001	−0.028 *
	(0.064)	(0.045)	(0.013)	(0.001)	(0.015)
No. of plots	−0.048	−0.038	0.011	0.001	−0.013
	(0.039)	(0.028)	(0.008)	(0.000)	(0.010)
Rented land	0.118	0.038	−0.012	−0.001	0.013
	(0.122)	(0.078)	(0.024)	(0.001)	(0.027)
Subsidy	0.002	0.003	−0.001	−0.000	0.000
	(0.007)	(0.004)	(0.001)	(0.000)	(0.000)
Off farm	0.495	0.318	−0.096	−0.005	0.109
	(0.328)	(0.224)	(0.068)	(0.004)	(0.076)
Cash deposit	0.386 *	0.281 *	−0.085 *	−0.004	0.096 *
	(0.222)	(0.154)	(0.047)	(0.003)	(0.052)
Camera monitor	0.001	0.011	−0.003	−0.000	0.004
	(0.067)	(0.043)	(0.013)	(0.001)	(0.015)
Constant	3.536 ***
	(0.898)
Observation	375	375	375	375	375

Note: ***, **, and * indicate statistical significance at 1%, 5% and 10%, respectively. The values in parentheses are robust standard errors.

. In the OLS regression results, the estimated coefficient is 0.448 and is significantly positive at the 5% level, indicating that farmers who have participated in the training have a higher willingness to adopt straw returning. Further, we used the ordered probit model for the regressions, and the estimated coefficient of the core explanatory variable—training—was positive and passed the significance test at the 1% level, after controlling for farmer characteristics, household characteristics, and village and community environment. In terms of marginal effects, we present the marginal effects when farmers’ willingness to adopt is 1, 3, and 5, respectively. When all covariates are at the mean, the probability that farmers feel “very much approved” after participating in the training increases by 0.126, the probability that farmers feel “average” decreases by 0.005, and the probability that farmers feel “very little approved” decreases by 0.111.This indicates that there is a significant positive effect of straw-returning technology training on farmers’ willingness to adopt the technology, i.e., technology training is beneficial to farmers’ willingness to adopt the technology. The findings of the study are similar to those of the literature that has studied the relationship between technology training and farmers’ technology adoption [29,30,31,32]. However, farmers’ self-learning variables did not affect their willingness to adopt straw-returning technology in either the OLS model or the ordered probit model. Therefore, why does joining the micro-groups of agricultural technology stations not affect farmers’ technology adoption intention? According to the theoretical analysis of the paper, we believe that farmers’ technology learning through WeChat technology sharing groups did not improve their cognition of the usefulness of technology and did not achieve better self-learning effects. There are two possible reasons for this effect on the learning of farmers: on the one hand, it may be because farmers are generally not well educated, and it is difficult to achieve a better learning effect through self-learning. On the other hand, it may be that farmers rely more on formal technical training than informal online self-learning, and they do not use the technical sharing groups on WeChat as their main source of new technical knowledge. We will subsequently test the above conjectures.

4.2. Robustness Check

4.2.1. Test of Endogeneity

It should be noted that when the ordered probit model was chosen to analyze the effect of training on farmers’ willingness to adopt technology, there may be endogeneity problems, such as omitted variables, two-way causality, and sample selection bias, resulting in inaccurate estimation results. Since the dependent variable willingness to adopt is an ordered variable, a direct regression using the Iv-probit model is not applicable. In order to overcome the potential endogenous problem, this paper refers to [33] and adopts the estimation method of conditional mixed process (CMP) for empirical testing, which combines the instrumental variables and the CMP estimation method and can better solve the endogeneity problem existing in the model.

Referring to the design idea of the instrumental variable in Zhang et al., (2019) [34], the mean value of farmers’ participation in technical training at the village level was selected as the instrumental variable in this paper, and the higher the proportion of farmers’ participation in training at the village level, the higher the likelihood of farmers’ participation in training, while the proportion of farmers’ participation in training at the village level can be considered an exogenous variable that does not affect farmers’ individual willingness to adopt straw return. To further test the validity of the instrumental variables, this paper refers to Chyi and Mao (2012) and uses multiple statistical tests of linear models to determine whether the selection of instrumental variables is valid [35]. The KleibergenPaap rk LM statistic is used to test whether the unincluded instrumental variables are correlated with the endogenous explanatory variables, and the original hypothesis of “under-identified instrumental variables” is significantly rejected at the 1% level. For the weak instrumental variables test, the KleibergenPaap rk Wald F-statistic of 61.065 is greater than the Stock–Yogo test 10% level threshold of 16.38, and the Cragg–Donald Wald F-statistic of 72.122 is greater than the Stock–Yogo test 15% level threshold of 8.96. Because only one instrumental variable is selected in this paper, it belongs to just the other case. Based on the above analysis, we believe that the mean value of farm household participation in technical training at the village level was a valid selection for an instrumental variable.

Columns (1) and (2) in

Table 4. Addressing endogeneity: instrumental variable estimation.

	IV-2SLS		CMP
Variables	(1)	(2)	(3)	(4)
	Technical Training Participation	Willingness	Technical Training Participation	Willingness
Technical training		0.849 *		0.669 *
		(0.471)		(0.370)
Instrumental variable	0.891 ***		3.116 ***
	(0.114)		(0.368)
Constant	0.023	2.929 ***		0.439 ***
	(0.174)	(0.746)		(0.037)
Control variable	Yes	Yes	Yes	Yes
atanhrho_12				0.110 **
				(0.053)
Observation	375	375	375	375
Under identification		0.000
Weak IV		61.065

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

report the estimated results of IV-2SLS, where training has a significant positive effect on farmers’ willingness to adopt the technology, which is significant at the 10% statistical level. This indicates that participating farmers are significantly more likely to have a very willing level of adoption than those who are not trained, i.e., training is more conducive to promoting the adoption of arable land conservation technologies by farmers. Columns (3) to (4) of Table 4 report the results of the CMP estimation with technology adoption willingness as the dependent variable. The estimated coefficients of training show that participation in training on straw-return technology significantly increases farmers’ willingness to adopt the technology, indicating that the regression results are more robust.

4.2.2. Examination of Other Self-Learning Pathways

Considering the reality that WeChat has become more popular in China’s rural areas in recent years, in the underlying regression model we use whether farmers join agricultural technology sharing groups on WeChat as a measure of farmers’ self-learning. However, whether farmers join agricultural technology sharing groups on WeChat is also influenced by factors such as whether they have smartphones and whether they can use smartphones skillfully. In the questionnaire, we recorded a variety of agricultural information acquisition channels for farmers, as follows: (1) whether farmers learn about agricultural technology through bulletin boards, (2) whether farmers often learn agricultural technology through the Internet, (3) whether farmers open the agricultural information SMS service (a kind of paid service that pushes agricultural information to farmers regularly through cell phone SMS), and (4) whether farmers learn agricultural technology on TV.

We used the above learning paths as proxy variables for farmers’ self-learning to test the robustness of the conclusion that self-learning does not affect farmers’ willingness to adopt straw-returning technologies. The test results are shown in

Table 5. Impact of self-learning on farmers’ willingness to adopt straw returning.

	Willingness to Adopt Straw Returning
Variables	(1)	(2)	(3)	(4)
Technical training	0.324 **	0.312 **	0.376 ***	0.368 **
	(0.144)	(0.142)	(0.144)	(0.150)
bulletin board	0.051
	(0.036)
Television		0.153
		(0.139)
Internet			0.023
			(0.059)
Message				0.020
				(0.168)
Control variable	Yes	Yes	Yes	Yes
Observation	375	375	375	375

Note: *** and ** indicate statistical significance at 1% and 5%, respectively.

, which shows that although farmers’ participation in the above four different self-learning approaches had a positive effect on their willingness to adopt straw returning technologies, none of them was significant. This indicates that our results are robust.

4.2.3. Replacement of Empirical Methods

Considering that the dependent variable of this paper is a dummy variable, we chose the propensity score matching method to test the robustness of the relationship between technical training and farmers’ willingness to adopt straw-returning technology by referring to Dillon (2011) and Wonde et al., (2022) [36,37].

The results of the propensity score matching method are shown in

Table 6. Impact of Training on farmers’ willingness to adopt straw returning: PSM method.

	(1)	(2)	(3)
	Kernel	Nearest Neighbor (1:5)	Caliper
treatment group	3.412	3.412	3.396
control group	3.007	2.970	3.001
ATT	0.405 *	0.442 *	0.395 *
	(0.2297)	(0.2452)	(0.224)
Observations	375	375	375

Note: * indicate statistical significance at 10%. Standard errors were calculated by using bootstrap method with repeat sampling 500 times.

. We chose three matching methods, namely kernel matching, 1:5 nearest neighbor matching and caliper matching, and the calculation results showed that farmers who attended technology training had a stronger willingness to adopt straw-returning technology under the three different matching modes, and it was significantly positive at the 10% level.

4.3. Mechanism Analysis

Based on the findings of the above analysis, why does training promote farmers’ willingness to adopt straw return? In this paper, we answer these questions from the perspective of improving farmers’ technology perceptions. In the questionnaire, we investigated farmers’ perceptions of straw-return technology and environmental protection. Firstly, “straw-return technology is more mature”, secondly, “do you think straw return has improved soil quality”, and thirdly, “straw return is beneficial to crop growth”, and fourth, “straw burning will bring serious air pollution problems” were measured using the five-point Likert method. We used these four questions to measure farmers’ perceptions of straw return and the environment. For both types of cognitive issues, we conducted empirical tests using technical training and farmers’ cognitive variables, and the regression results are shown in

Table 7. Mechanism analysis of technical training affecting farmers’ adoption willingness.

Variables	(1)	(2)	(3)	(4)
	Environmental Cognition	Technical Cognitive
	Air Quality	Crop Growth	Technical Maturity	Land Quality
Technical training	0.198	0.317 ***	0.401 ***	0.324 **
	(0.162)	(0.112)	(0.137)	(0.144)
Self-learning	0.155	0.336	0.264	0.441
	(0.321)	(0.260)	(0.258)	(0.282)
Control variable	Yes	Yes	Yes	Yes
Observation	375	375	375	375

Note: *** and ** indicate statistical significance at 1% and 5%, respectively.

. The results of the empirical test found that formal technology training only reinforced farmers’ perceptions of the straw-returning technology, and did not have an effect on farmers’ environmental perceptions. This finding is consistent with the hypothesis that farmers are economically rational human beings and that whether farmers adopt a new technology depends on the expected benefits of adopting the technology, and whether it contributes to environmental improvements with externalities is not a priority for farmers.

The regression results in columns (2) to (4) in Table 7 show that farmers who have participated in technology training have a higher level of awareness of the usefulness of the technology in three ways: first, they recognize that the straw-return technology is more mature and can be adopted with confidence. Second, they are more likely to recognize that the straw-return technology can help promote the improvement of arable land quality, and third, they believe that the adoption of the straw-return technology can help the growth of crops. It has been found that in order to maximize utility, farmers adjust their land-use decisions according to their different perceptions of arable land quality conservation [38], and when the level of farmers’ perceptions of arable land conservation technologies increases, it helps to promote farmers’ willingness or behavior to adopt the relevant technology [28,39]. Therefore, we found that formal technology training promotes farmers’ willingness to adopt the technology by increasing their perception of the usefulness of straw-return technology.

At the same time, we also found that whether or not farmers participated in the WeChat technology sharing group did not have a significant effect on the cognition of straw-returning technology, which to some extent indicates the poor learning effect of farmers’ participation in the technology sharing group. We will further explore in the heterogeneity analysis whether the farmers’ education level is a possible reason for the effect of farmers’ self-learning.

4.4. Heterogeneity Analysis

Based on the above analysis, it is clear that straw returning affects farmers’ willingness to adopt technology by improving their knowledge, so that the effect of technical training is closely related to farmers’ knowledge absorption ability. For farmers with different abilities, training may have different effects, e.g., for farmers with better knowledge absorption or learning ability, training may be more likely to increase their knowledge of straw-returning technology and promote their willingness to adopt it. The results in columns (1) and (2) of

Table 8. Heterogeneous effects of technology training and self-learning on the willingness of different types of farmers to adopt technology.

Variable	(1)	(2)	(3)	(4)
	High Level of Education	Low Level of Education	WeChat Group	Non-WeChat Group
Technical training	0.376 *	0.246	0.821 **	0.288 *
	(0.212)	(0.214)	(0.387)	(0.165)
Self-learning	0.360	−0.152
	(0.336)	(0.453)
Control variable	Yes	Yes	Yes	Yes
Observation	176	199	56	319

Note: ** and * indicate statistical significance at 5% and 10%, respectively.

show that training only promotes the adoption willingness of farmers with higher education levels, and does not have a significant effect on the adoption intention of farmers with lower education levels.

At the same time, we did not find that the participation of farmers with higher education level in the WeChat technology training group had a significant effect on their willingness to adopt the straw-returning technology, and we can say that the level of farmers’ learning ability is not the reason for the failure of the technology sharing group to play a role in technology promotion. In China, farmers rely mainly on government training (Li et al., 2017) and joining cooperatives (Wan et al., 2021) to acquire new agricultural technologies [40,41]. In the above-mentioned ways of acquiring new technologies, farmers receive detailed technical explanations and operational demonstrations, which can better reduce farmers’ perceived risks and uncertainties about new technologies. Although the information technology that has emerged in recent years can reduce the cost of farmers’ access to information, it is difficult to provide farmers with detailed technical explanations and answer their doubts about the cognition of new technologies, so its effect on their understanding of new technologies is limited.

Further, we distinguish farmers into two categories according to whether they join the WeChat group or not, and the results of the empirical tests are shown in columns (5) and (6) of Table 6. It was found that the technical training helped to improve farmers’ willingness to adopt the straw-returning technology regardless of whether they joined the WeChat group or not, but for farmers who joined the WeChat group, the coefficient of influence of technical training on their willingness to adopt was 0.821, which was significantly positive at the 5% level, and the technical training had a stronger effect on the willingness to adopt the technology for farmers who joined the WeChat group. In other words, although joining the WeChat group does not directly affect farmers’ technology adoption, it can serve as a useful supplement to the formal training system.

5. Conclusions and Policy Implications

Farmers are the key actors in the protection of arable land quality, and the question of how to promote farmers’ willingness to adopt conservation technologies is the focus of this paper. Based on this, this paper explores the key factors that influence farmers’ willingness to adopt straw-returning technologies from the perspective of formal technical training and informal online self-directed learning. The results of this study show that farmers who participated in straw-returning technology training significantly increased their willingness to adopt the technology, but participation in online self-directed learning did not affect farmers’ willingness to adopt straw returning. These findings hold true after overcoming endogeneity. The mechanism analysis found that farmers who participated in the technology training significantly increased their perceived usefulness of straw-returning technology, which in turn improved their willingness to adopt straw returning. However, it did not have a significant effect on farmers’ perceptions of their straw burning environment. Further analysis revealed that technology training only promoted adoption willingness among better-educated farmers. For farmers who did not participate in the technology sharing group, the willingness to return straw to the field was higher for farmers who participated in both the technology training and the WeChat technology sharing group.

Based on the above findings, this paper makes the following policy recommendations: First, increase technical training for farmers, with large-scale growers and new farmers as the core training subjects. Second, further enrich the training content on the usefulness of arable land conservation technology and strengthen farmers’ knowledge of arable land conservation technology. Third, actively use information technology to build a diversified agricultural technology training system with formal training as the main body and information technology training as the supplement.