The Effect of Agricultural Extension Service Need-Supply Fit on Biological Pesticides Adoption Behavior: Evidence from Chinese Rice Farmers

Abstract

Agricultural extension services play an important role in promoting pesticide reduction and green production, although the mismatch between farmers’ needs and service supply in rural China seriously affects the application and promotion of biological pesticides, which has been generally ignored by existing studies. Taking 1160 rice farmers in Hubei Province, China, as an example, this study constructs a need-supply fit model of agricultural extension services from the perspectives of need and supply. We further use the logit model and propensity score matching method (PSM) to estimate the effect of service need-supply fit on farmers’ biological pesticide adoption behavior. The specific results are as follows: First, the level of need-supply fit for agricultural extension services was low for the overall sample. This indicates that the agricultural extension service supply is not consistent with the actual needs of farmers. Second, the agricultural extension service need-supply fit significantly and positively affects farmers’ biological pesticide adoption behavior. Third, there are significant differences in farmers’ biological pesticide adoption behaviors under different need-supply fit. When the level of service need-supply fit exceeds the 0.2 threshold, the positive effect of service need-supply fit on farmers’ biological pesticide adoption behavior gradually increases. Fourth, the agricultural extension service need-supply fit indirectly increases farmers’ motivation to adopt biological pesticides by regularizing biological pesticide use behavior and alleviating path dependency. Therefore, the main body of agricultural extension services should optimize the adjustment of the supply mode of agricultural extension services to fully grasp the reality of farmers’ need for biological pesticides in order to promote the application of biological pesticides.

1. Introduction

According to statistics, China has one of the highest levels of pesticide use in the world, with an average pesticide use per ha of approximately 1.5 to 4 times higher than the world average [1]. For this reason, the lack of non-chemical alternatives has been a major factor preventing farmers from reducing pesticide use [2]. Biological pesticides have long been promoted as potential alternatives to chemical pesticides. And it has also been considered safe for both humans and the environment [3]. Biological pesticides are natural substances derived from animals, plants, bacteria, fungi, viruses, and several minerals. They have the advantages of low toxicity, rapid decomposition, low residues, and high targeting [4]. Therefore, it is clear that compared with chemical pesticides, biological pesticides have more advantages in controlling pests and diseases. However, as of December 2021, there were 141 registered active ingredients of biological pesticides and 2105 biological pesticide products in China (excluding agricultural antibiotics and natural enemies) [5]. Although China’s annual production of biological pesticides reaches 14 million tons, with an annual output value of about 3 billion yuan (Yuan is the unit of currency in China. The exchange rate was 7.285 yuan to one dollar in August 2023. The exchange rate is the exchange rate information from the Bank of China), this only accounts for about 10% of the overall total pesticide value [6]. At the same time, some data show that China’s application ratio of biological pesticides is only 12.10% [7]. Then, why is the adoption rate of biological pesticides so poor despite their better substitution characteristics? It has been shown that incomplete information on biological pesticides among farmers [8], risk preference [9], lower extent of technical training [10], higher cost of biological pesticides [11], delayed effect of biological pesticide technology, limited marketing and sales channels [8], and an inadequate legal system [12] have seriously hindered the adoption of biological pesticides by farmers. Therefore, how do we encourage farmers to adopt biological pesticides?

Agricultural extension services provide a potential solution. As an essential way to help farmers master or adopt green production methods, agricultural extension services can deliver information about new technologies to farmers, thus promoting the adoption of new technologies [13]. It includes not only promoting new technologies but also educating farmers, training farmers, and improving farmers’ practical lives [14]. Agricultural extension services have the following two advantages in promoting farmers’ new technologies adoption. A part of the literature found a positive role for agricultural extension services in the diffusion of new technologies from the level of type and intensity of agricultural extension. For example, Liu et al. [6] found that technical training had a significant positive effect on the adoption of biological pesticides by farmers. And the more training courses farmers participated in, the higher the probability of farmers adopting biological pesticides. Niyaki et al. [15] revealed that farmers’ participation frequency in education and extension activities showed a significant positive effect on their use of biological control of Trichogramma. Ommani et al. [16] suggested that extension activities, social participation, and information access presented a significant positive effect on the farmers’ use of biological control of Bracon parasitoid. The other part of the literature analyzed the level of farmers’ needs and service supply and found that the mismatch between service need and supply seriously weakened the effectiveness of agricultural extension services. For example, Dominik et al. [17] showed that farmers would participate in agricultural extension only if the services and information they demanded were specifically recommended and answered. Rebuffel et al. [18] studied French farmers and found that local advisory services supply only a small proportion of the population while neglecting to reach out to other different types of farmers.

Usually, agricultural extension service organizations regularly deliver uniform extension content and services to farmers [19]. Although most farmers receive services about biological pesticides, it is difficult for farmers to express in their own words or lack channels to express whether they need the service [20,21]. Consequently, the delivered information may not meet the needs of the farmers and may lead to an unmatched condition of consistency between the agricultural extension services supply and the farmers’ needs. This will not only reduce the enthusiasm of farmers to participate in agricultural extension but also generate lower effectiveness of agricultural extension services, decrease efficiency of pest control, and waste public resources [17,22]. Thus, it is necessary to measure the need-supply fit of agricultural extension services from the perspective of matching need and supply and discuss whether the service need-supply fit affects the farmers’ adoption of biological pesticides.

In conclusion, our study attempts to answer the following three questions. What is the match between farmers’ actual needs and agricultural extension service supply? Does the agricultural extension service need-supply fit enhance farmers’ biological pesticide adoption? How does agricultural extension service need-supply fit affect the adoption of biological pesticides by farmers? China has a long history of rice cultivation, and rice is one of the most important food crops. During rice cultivation, chemical pesticides are used in excess and inefficiently, which can lead to problems of soil caking and surface source pollution [23]. Therefore, this paper presented research on pesticide inputs in rice production. On this basis, using survey data from 1160 rice farmers in Hubei Province, China, this study uses a need-supply fit model to construct an agricultural extension service need-supply fit framework for farmers. It discusses the effect of agricultural extension service need-supply fit on farmers’ biological pesticide adoption and research pathways. In particular, the differences in farmers’ biological pesticide adoption behavior under different need-supply fit states are verified. Distinguishing from existing studies, this study may have the following three contributions.

Firstly, studies have focused on whether farmers participate in agricultural extension services and the frequency of their participation. However, the importance of the content of agricultural extension services has been overlooked. Agricultural extension service content is a decisive factor affecting farmers’ participation in agricultural extension and adopting new agricultural technologies. Only when agricultural extension content satisfies farmers’ needs will farmers actively accept agricultural extension services. Thus, this study takes agricultural extension services content as the research perspective. Drawing on Mkenda et al. [10], agricultural extension services are discussed in terms of five aspects: pest and disease warning service, biological pesticide service, pesticide dosage guidance service, pesticide residue testing service, and marketing service.

Secondly, most studies have discussed how to improve the effectiveness of agricultural extension services from a single dimension or subject. However, it ignores the important role of matching needs and supply in improving the effectiveness of agricultural extension services. Therefore, with the help of the need-supply fit model, this study measures the level of fit between agricultural extension farmers’ needs and service supply and examines the consistency of agricultural extension service supply and farmers’ needs. Then, the factors influencing farmers’ need-supply fit are discussed.

Thirdly, studies have discussed biological pesticide adoption by farmers in terms of internal and external factors. However, the important influence of agricultural extension service need-supply fit on farmers’ adoption behavior has been ignored. Based on this, the present study discusses the direct effects and indirect pathways of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior from two perspectives: biological pesticide use regularization behavior and path dependence. It further tests that there are significant differences between different degrees of services need-supply fit on farmers’ biological pesticide adoption behavior.

2. Theoretical Analysis and Research Hypotheses

2.1. Impact of Agricultural Extension Service Need-Supply Fit on Farmers’ Biological Pesticide Adoption Behavior

Agricultural extension services perform an essential role in facilitating the adoption of new technologies by farmers. However, the low level of information available to farmers on advanced technologies can lead to significant differences in adoption decisions [13]. The theory of technology supply and need suggests that the mismatch between the technological information needed by farmers and the content promoted by service providers will affect the application and diffusion of new technologies [24]. Therefore, the level of need-supply fit of agricultural extension services will directly affect farmers’ biological pesticide adoption behavior. When the level of service need-supply fit is higher, the services that farmers need are close to the same as the services provided by the service main body, which not only reduces the learning cost of biological pesticide technology for farmers [25] but also solves the problem of resource wastage in the process of technology promotion. Due to their characteristics, biological pesticides cannot be mixed with alkaline drugs (Bordeaux solution) or fungicides (Tricyclazole) [26]. If farmers ignore this, it may reduce the effectiveness of rice control and yield. Taking the pesticide dosage guidance service as an example, the agricultural extension service subject provides farmers with services such as pesticide dosage guidance and mixing and proportioning guidance to reduce the uncertainty of the operation of individual biological pesticide technology [11,27]. Also, in order to guarantee the effective promotion of biological pesticides, it will help farmers solve the difficulties encountered in controlling pests and diseases in the field. In addition, when farmers’ technology needs were addressed, farmers’ costs of information and time to access biological pesticides were reduced, further increasing their likelihood of adopting biological pesticides [25]. Studies have shown that farmers’ decision-making behavior on biological pesticides is inhibited when their effective needs for biological pesticides are not addressed [28]. In addition, Luan et al. [29], citing agricultural productive services as an example, found that a higher level of service need-supply fit can achieve the division of labor in agriculture and promote the optimal allocation of factors, thereby promoting the enthusiasm of farmers to engage in agricultural production. Based on the above discussion, Hypothesis 1 is proposed:

Hypothesis 1.

Agricultural extension service need-supply fit significantly promotes farmers’ biological pesticide adoption behavior.

2.2. Analysis of Research Pathways of Agricultural Extension Service Need-Supply Fit on Farmers’ Biological Pesticide Adoption Behavior

Service need-supply fit is an optimal status in the process of agricultural technology diffusion, and the following two advantages exist in promoting the adoption of biological pesticides by farmers. One of them is to regulate the use of farmers’ biological pesticides. The long-term unregulated dosing behavior is an important cause of pesticide overuse [30,31]. Nevertheless, agricultural extension services provide farmers with accurate pesticide spraying times, precise pesticide proportions, pest control information, and proper biological pesticide products through pest and disease early warning services, pesticide dosing guidance services, and biological pesticide services. They are able to effectively rectify irregular pesticide use behaviors by farmers [28]. Thus, the control effect of biological pesticides is optimized. The second is to alleviate the path dependence of farmers. The farmers who have been engaged in agricultural production for a long time have rich experience in cultivation. However, their knowledge system was relatively old, and their ability to learn and apply new technologies, such as biological pesticides, was relatively poor [32]. Also, farmers relying on production experience are more inclined to use traditional chemical pesticides, thus abandoning the use of biological pesticides [33]. While agricultural extension service providers deliver information on biological pesticide technology to farmers, they also improve farmers’ understanding of the ease of use of the technology [34], which can, to some extent, break farmers’ path dependence on traditional cultivation experience, thus increasing the likelihood of their adoption of biological pesticides. When the need and supply of services are matched, the frequency of communication and interaction between farmers and service providers is higher, which is conducive to farmers’ familiarity with the types of crop pests and diseases, the correct matching of types of pesticides, and the elimination of farmers’ concerns about the adoption of new technologies [3,35]. Thus, it drives changes in farmers’ decision-making on biological pesticide adoption. On the basis of the above discussion, Hypotheses 2 and Hypotheses 3 are proposed:

Hypotheses 2.

Agricultural extension service need-supply fit influences farmers’ biological pesticide adoption behavior by regularizing biological pesticide use behavior.

Hypotheses 3.

Agricultural extension service need-supply fit can effectively mitigate path dependence and lead to the adoption of biological pesticides by farmers.

3. Materials and Methods

3.1. Data Collection

In this study, the main rice production area in Hubei Province, China, was selected as the survey area, and microscopic research was conducted in 2021. As one of the major grain-producing areas in China, Hubei Province ranked 5th in the country in terms of rice sown area in both 2019 and 2020, accounting for 7.70% and 7.58%, respectively. Hubei Province was one of the provinces with the most pesticide use in China, ranking 4th in both 2018 and 2019, accounting for 6.87% and 6.97% of pesticide use, respectively [36]. Based on this, rice farmers in Hubei Province were selected as the research object, which is of some significance for studying pesticide reduction.

The sample sampling process for this study was as follows. First, we randomly selected 10 representative rice-producing counties in Hubei Province, including Qianjiang, Xiantao, Xiangzhou, Zaoyang, Shishou, Jianli, Tuanfeng, Xishui, Jingshan, Shayang. Second, we randomly selected three townships in each county. Third, three villages were randomly chosen in each town. Fourth, the sample of farm households was selected based on the list of farm households provided by the village managers and using an equidistant sampling method (equidistance of 10), with 13–14 farmers selected from each village. Finally, we obtained 1193 questionnaires. Then, after excluding the samples that had missing variables and inconsistent information, we had a total of 1160 valid samples, and the sample efficiency rate was 97.23%.

Before the formal research, we made the following preparations: first, in the selection of researchers. Master’s and doctoral students who are experienced in research were selected, and they were given professional training. Secondly, we conducted a preliminary survey and optimized and adjusted the content and structure of the research questionnaire based on the actual research situation. The research questionnaire mainly included the following six aspects: farmer characteristics, family characteristics, farmland characteristics, biological pesticide extension service, pesticide and fertilizer use, and infrastructure.

3.2. Variable Definitions and Econometric Model

3.2.1. Variable Definitions

(1)

Explained variable

The explained variable is the farmer’s biological pesticide adoption behavior. Drawing on the research of Huang et al. [28] and Tang et al. [37], the farmer’s biological pesticide adoption behavior is defined as “Do you adopt biological pesticide? If “yes,” the value is assigned as “1,” otherwise “0”.

(2)

Explanatory variable

The explanatory variable is service need-supply fit. In this study, service need-supply fit refers to the consistency or degree of fit between the service content provided by the agricultural extension service and the service needs or preferences of the farmers. The fit model originates from interactive psychology, also known as “individual-organizational” fit theory, which refers to a state of mutual equilibrium or the degree of overall fit [38,39,40]. This theory has now been widely used in the fields of economics and management. The “need-supply” fit refers to the match between human need and environmental supply, reflecting a balance between need and supply [41]. Given this, drawing on Kristof [41] and Cable et al. [42], this study constructs a need-supply fit model for agricultural extension services from the need and supply perspectives. To reflect the level of consistent match between the content of agricultural extension services supply and farmers’ needs [43]. The following are the steps for measuring the need-supply fit of agricultural extension services for farmers:

Firstly, drawing on Mkenda et al. [10], agricultural extension services are divided into five aspects: pest and disease warning service, biological pesticide service, pesticide dosage guidance service, pesticide residue testing service, and marketing service.

Secondly, we present separately the need and supply information of different farmers regarding the five types of agricultural extension services. If farmers have a need for a type of service, it is assigned a value of “1”; otherwise, it is “0”. If the local agricultural extension department provides one of the above services, it means that there is a local supply of that type of service, which is assigned a value of 1; otherwise, it is assigned a value of 0. From the perspective of the permutation approach, there are four different scenarios of matching need and supply for agricultural extension services, i.e., “Need = 0, Supply = 1”, “Need = 1, Supply = 1”, “Need = 1, Supply = 0”, “Need = 0, Supply = 0”. There is only one algorithm specified here, in which the final need-supply fit is 1 only if the need and supply assignments are both 1, and it is 0 in all other cases. In other words, only when there is a combination of “Need = 1, Supply = 1” in the matching of need and supply of agricultural extension services is the service considered to be in need-supply fit.

Table 1. Status of value assignment of different agricultural extension service supply and farmers’ need.

Agricultural Extension Service		Need = 0, Supply = 1	Need = 1, Supply = 1	Need = 1, Supply = 0	Need = 0, Supply = 0
Pest and disease warning service	$q_{1i}$	(0, 1)	(1,1 )	(1, 0)	(0, 0)
Biological pesticide service	$q_{2i}$	(0, 1)	(1, 1)	(1, 0)	(0, 0)
Pesticide dosage guidance service	$q_{3i}$	(0, 1)	(1, 1)	(1, 0)	(0, 0)
Pesticide residue testing service	$q_{4i}$	(0, 1)	(1, 1)	(1, 0)	(0, 0)
Marketing service	$q_{5i}$	(0, 1)	(1, 1)	(1, 0)	(0, 0)

Note: When the need and supply states are (0,1) (1,0), and (0,0) are assigned the value “0”; when it is (1,1,) is assigned the value “1”.

presents the combination of need and supply for five types of agricultural extension services. The model is constructed as follows:

$$q_i=q_{1i}+q_{2i}+q_{3i}+q_{4i}+q_{5i}$$

(1)

In Equation (1), $q_i$ is the need-supply fit of the five agricultural extension services of the $i$th farmers. $q_{1i}$ is the need-supply fit of the pest and disease warning service of the $i$th farmers; $q_{2i}$ is the need-supply fit of the biological pesticide service of the $i$th farmers; $q_{3i}$ is the need-supply fit of the pesticide dosage guidance service of the $i$th farmers; $q_{4i}$ is the need-supply fit of the pesticide residue testing service of the $i$th farmers; $q_{5i}$ is the need-supply fit of the marketing service of the $i$th farmer.

Third, the need-supply fit of agricultural extension services for each farmer was calculated. The sum of the five extension services in the “Need = 1, Supply = 1” state for each farmer and the weight of the five agricultural extension services for that farm. A higher level of need-supply fit for farmers’ services indicates a better matching between the need for and supply of agricultural extension services. The model is constructed as follows:

$$Z_i=\frac{q_i}{n}$$

(2)

In Equation (2), $Z_i$ is the level of need-supply fit of agricultural extension services for the $i$th farmer; $n$ is the number of agricultural extension service components, i.e., it takes the value of 5.

(3)

Control variables

Drawing on existing studies, this study selected control variables from six aspects, including farmer characteristics (age, gender, health, education level, part-off farm) [37,44], family characteristics (family income, family size) [15,44], farmland characteristics (farmland size, the quality of farmland) [16,28], infrastructure (internet use, distance to market) [3,27,45], technical training [27,46,47], and region dummy variables.

(4)

Mechanism variables

The mechanism variables are biological pesticide use regularization and path dependence. First, drawing on zhang et al. [47], whether farmers follow safety intervals when using biological pesticides is defined as regularizing biological pesticide use behavior. The specific question is, “Do you comply with the safety interval when using biological pesticide? If yes, it is assigned a value of 1, otherwise it is 0”. Secondly, path dependence refers to the dependence of farmers on their past experience of pesticide use for biological pesticides. The specific question is, “Do you rely on personal experience when using biological pesticide? If yes, it is assigned a value of 1, otherwise it is 0”.

3.2.2. Econometric Model

This study focused on the effect of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior. In this study, the explained variable is the farmers’ biological pesticide adoption behavior, and it is a binary discrete variable, so we use a binary logit model to estimate the sample. The model is set as follows.

$$\ln (\frac{P}{1-P})={\beta }_0+{\beta }_1{Z}_i+{\displaystyle {\sum }_{i=1}^n{\beta }_i}{X}_i+\epsilon$$

(3)

In Equation (3), $P$ is the probability of a farmer adopting a biological pesticide. $Z_i$ is the core explanatory variable (service need-supply fit). $X_i$ is the control variable. ${\beta }_1$ is the regression coefficient of the core explanatory variable. ${\beta }_i$ is the regression coefficient of the control variable. ${\beta }_0$ is the intercept of the regression. $\epsilon$ is the random error term.

Table 2. Descriptive statistics.

Variable		Mean	Standard Deviation	Minimum	Maximum	Sample Size
Biological pesticide adoption behavior	1 if the farmer adopts a biological pesticide; 0 otherwise	0.714	0.452	0	1	1160
Biological pesticide use intensity	Cost of biological pesticides as a percentage of total pesticide costs	0.430	0.360	0	1	1160
Service need-supply fit	See variable definitions for the calculation procedure	0.373	0.293	0	1	1160
Number of supply and need imbalance	See variable definitions for the calculation procedure	1.781	1.299	0	5	1160
Age	Age (years)	59.365	9.026	25	78	1160
Gender	1 if male; 0 if female	0.820	0.384	0	1	1160
Health	1 if the farmer has a chronic disease; * 0 otherwise	0.295	0.456	0	1	1160
Education level	Education years (years)	7.412	3.449	0	16	1160
Part-off work	Total part-off work hours (months)	1.511	3.252	0	12	1160
Family income	Family income (ten thousand yuan $)	10.314	21.478	0.0354	303.7	1160
Family size	Number of family size (person)	2.818	1.288	0	9	1160
Farmland size	The size of farmland (in logarithmic)	2.409	1.169	−0.916	7.601	1160
Quality of farmland	poor = 1; common = 2; better = 3	2.247	0.699	1	3	1160
Distance to market	Distance to market (in kilometers)	3.571	5.034	0	120	1160
Internet use	1 if the farmer uses the internet; 0 otherwise	0.549	0.498	0	1	1160
Technical training	1 if the farmer participates in technical training; 0 otherwise	0.239	0.427	0	1	1160
Biological pesticide use regularization	1 if the farmer complies with the safety interval when using biological pesticide; 0 otherwise	0.666	0.472	0	1	1160
Path dependence	1 if the farmer relies on personal experience by using biological pesticide; 0 otherwise	0.323	0.468	0	1	1160
Jing men	1 if Jing men; 0 otherwise	0.209	0.406	0	1	1160
Jing zhou	1 if Jing zhou; 0 otherwise	0.195	0.396	0	1	1160
Xiang yang	1 if Xiang yang; 0 otherwise	0.232	0.422	0	1	1160
Huang gang	1 if Huang gang; 0 otherwise	0.190	0.392	0	1	1160

Note: * Common chronic diseases include cancer, diabetes, respiratory chronic diseases, and cardiovascular diseases, with cardiovascular diseases including heart disease and coronary heart disease. ^$ yuan is the Chinese currency unit, which was 7.285 yuan to one dollar in August 2023. The exchange rate is the exchange rate information from the Bank of China.

reports the descriptive statistical analysis of the variables.

3.2.3. Propensity Score Matching Method

Although the previous section estimated the effect of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior, the model estimates may be biased due to some uncontrollable factors. Therefore, we use the propensity score matching method (PSM) to address the endogeneity problem that exists in the sample. It should be emphasized that since the treatment variable of the article, service need-supply fit, is a continuous variable. Drawing on Yan et al. [46], we define service need-supply fit as “low-fit” if it is less than the mean and “high-fit” if it is greater than the mean. The propensity score matching method aims to match sample farmers in the control group (low-fit) with sample farmers in the treatment group (high-fit). All other things being equal, the causal relationship between the need-supply fit of services on farmers’ biological pesticide adoption behavior was determined by the differences in farmers’ biological pesticide use between the control and treatment groups. The propensity score matching method (PSM) model consists of the following three steps.

First, a logit regression is used to estimate the propensity score value for services need-supply fit as follows:

$$P(X_i)=P(Z=1|X_i)=\exp (X_i^{\prime }\beta )/(1+\exp (X_i^{\prime }\beta ))$$

(4)

In Equation (4) $P(Z=1|X_i)$ is the propensity score for service need-supply fit. $X_i$ is a covariate variable. When covariates are considered, the variable has an impact on both the service need-supply fit and on farmers’ biological pesticide adoption behavior. The covariates selected for this study include farmer characteristics (age, gender, health, education level, part-off farm), family characteristics (family income, family size), farmland characteristics (farmland size, the quality of farmland), infrastructure (internet use, distance to market), technical training, and region dummy variables.

Second, this study used different matching methods to match the sample farmers in the control group (low-fit) with the sample farmers in the treatment group (high-fit) and performed a balancing test to ensure the comparability of the control and treatment groups.

Third, the average treatment effect of the treatment (ATT) was calculated. After matching is completed, the effect of service need-supply fit on farmers’ biological pesticide adoption behavior is measured. The specific expressions are as follows:

$$ATT=E(Y_1|Z=1)-E(Y_0|Z=1)=E(Y_1-Y_0|Z=1)$$

(5)

In Equation (5), $Y_1$ and $Y_0$ represent the two counterfactual results for the control group (low-fit) and the treatment group (high-fit), respectively.

4. Results and Discussion

4.1. The Status of the Farmers’ Need and Agricultural Extension Service Supply

In this study, we first analyze the basic status of farmers’ needs and agricultural extension service supply (see

Table 3. Analysis of need-supply fit statistics of different agricultural extension services.

	Pest and Disease Warning Service	Biological Pesticide Service	Pesticide Dosage Guidance Service	Pesticide Residue Testing Service	Marketing Service
Supply	901	700	1013	108	206
Need	848	701	845	511	537
The number of service need-supply fit	661	526	764	61	132
Service need-supply fit	56.98%	45.34%	65.86%	5.26%	11.38%
Total service need-supply fit level	36.97%

). It can be seen that in the pesticide dosage guidance service group, 1013 farmers received this service from the agricultural extension service organization, while the sample that had a need for this service was 845. The percentage of samples that reached the need-supply fit was 65.86%. In the pest and disease warning service group, 901 farmers obtained the service from agricultural extension service organizations, while the sample with a need for the service consisted of 848 farmers. The proportion of samples that reached the need-supply fit was 56.98%. In the group providing biological pesticide services, 700 farmers learned about or got the service from the agricultural extension service organization, whereas the sample with the need for the service was 526 farmers. The percentage of samples that reached supply-demand fit was 45.34%. In addition, the need-supply fit level of pesticide residue testing service was less than 10%. It can be seen that the level of need-supply fit for the five types of agricultural extension services, in descending order, are pesticide dosage guidance services, pest and disease warning services, biological pesticide services, pesticide residue testing services, and marketing services. In summary, despite the high level of need and supply for each service, most of the supply of agricultural extension services does not actually meet the needs of farmers, which can lead to a reduction in the effectiveness of agricultural extension services.

4.2. Analysis of Influence of the Agricultural Extension Service Need-Supply Fit on Farmers’ Biological Pesticide Adoption Behavior

4.2.1. The Influence of Agricultural Extension Service Need-Supply Fit on Farmers’ Biological Pesticide Adoption Behavior

In this investigation, SPSS (19.0) is used to analyze the correlation between the agricultural extension service need-supply fit and farmers’ biological pesticide use behavior. The services need-supply fit is defined as “Low-fit“ if it is below the average value or “High-fit” if it is above the average value, based on the study of Tang et al. [37] and Yan et al. [46].

Table 4. Status of biological pesticide adoption by farmers with different states of need-supply fit.

	Not Use Biological Pesticides		Use Biological Pesticides
	Sample	Percentage	Sample	Percentage
Low-fit	174	40.65%	254	59.35%
High-fit	158	21.58%	574	78.42%
Pearson ${\chi }^2$	48.075 ***

Note: *** indicate significance at the 1% significance levels.

presents the current status of biological pesticide use by farmers under different need-supply fit states. The results show that the proportion of low-fit level farmers adopting biological pesticides is 59.35%, while the percentage of high-fit level farmers adopting biological pesticides is 78.42%. And the chi-square test was significant at a 1% statistical level. The results show that compared to farmers with low-fit levels, farmers with high-fit levels have a higher probability of using biological pesticides. And there is a significant difference in the adoption behavior of biological pesticides between farmers with low-fit and high-fit. This result tentatively verifies the higher probability of biological pesticide adoption among farmers with a high need-supply fit.

Considering the multicollinearity problem between variables, this study conducts multicollinearity tests for each explanatory variable (see

Table A1. Variance inflation factor.

	VIF	1/VIF
Service need-supply fit	1.05	0.952
Age	1.64	0.608
Gender	1.20	0.836
Health	1.06	0.942
Education level	1.24	0.807
Part-off work	1.12	0.892
Family income	1.22	0.822
Family size	1.03	0.968
Farmland size	1.37	0.732
Quality of farmland	1.04	0.963
Distance to market	1.02	0.979
Internet use	1.47	0.608
Technical training	1.14	0.876
Mean VIF	1.20	-

Abbreviation: VIF, variance inflation factor.

). The results show that the variance inflation factor (VIF) values of each explanatory variable are below 2. This indicates that there is no serious problem of multicollinearity among the explanatory variables in the sample. Based on this, this study further estimates the effect of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior (see

Table 5. Results of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior.

Variable	Model 1		Model 2		Model 3
	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error
Service need-supply fit	2.011 ***	0.242	2.024 ***	0.265	2.349 ***	0.350
Age			−0.001	0.010	0.019	0.012
Gender			0.240	0.194	0.572 **	0.247
Health			0.211	0.162	−0.059	0.203
Education level			0.026	0.023	0.024	0.027
Part-off work			0.033	0.026	−0.013	0.030
Family income			0.004	0.006	0.034 ***	0.012
Family size			−0.138 **	0.056	−0.174 ***	0.065
Farmland size			0.468 ***	0.074	0.147	0.097
Quality of farmland			0.141	0.104	0.164	0.124
Distance to market			0.107 ***	0.031	0.003	0.033
Internet use			0.267	0.174	0.477 **	0.205
Technical training			0.448 **	0.201	0.644 ***	0.231
Jing men					0.132	0.287
Jing zhou					−0.633 **	0.249
Xiang yang					4.189 ***	0.742
Huang gang					−2.167 ***	0.276
Constant	0.206 **	0.099	−1.750 **	0.77	−2.260 **	0.939
Pseudo R2	0.060 ***		0.159 ***		0.376 ***
LR Chi2	82.76		221.259		522.620

Note: **, *** indicate significance at the 5%, 1% significance levels, respectively. Note: Model 1 incorporated only the effects of core explanatory variables on farmers’ biological pesticide adoption behavior. Model 2 included the effects of core explanatory variables and control variables on farmers’ biological pesticide adoption behavior. Model 3 incorporated the effects of the core explanatory variables, control variables, and regional control variables on farmers’ biological pesticide adoption behavior.

). Model 1 includes only the regression results of the need-supply fit of agricultural extension services. It shows that the need-supply fit of agricultural extension services had a significant positive effect on farmers’ biological pesticide adoption behavior. Model 2 incorporates the regression results for all control variables. The results indicate that the need-supply fit of agricultural extension services remains significant in positively influencing farmers’ biological pesticide adoption behavior. Model 3 is the regression result after the inclusion of the regional control variable in Model 2, which is found to remain significant. The goodness of fit of Model 3 is better than both Model 1 and Model 2, so the results of Model 3 are analyzed. It was found that the more the level of services need-supply fit, the higher the probability of adoption of biological pesticides by farmers, and Hypothesis 1 was tested. The possible reason for this is a higher match between agricultural extension service provision and farmers’ real needs. This implies that the level of complete information on biological pesticides is well enhanced for farmers, which reduces their production and technology risks of using biological pesticides and thus increases the motivation to adopt biological pesticides [11,15].

In terms of control variables, men, as the main laborers in agricultural production, are more knowledgeable about pest and disease control in the field and are more inclined to proactively access resources and information about agriculture. Thus, male farmers are more motivated to use less toxic and efficient biological pesticides for pest and disease control [3].

There is a significant positive effect of family income on the biological pesticide adoption behavior of farmers. Compared to the price of chemical pesticides, the price of biological pesticides is generally higher. Farmers are more likely to adopt biological pesticides when they have sufficient income and initial assets [16]. Family size shows a significant negative effect on farmers’ biological pesticide adoption behavior. When households have a larger number of laborers, they are able to alleviate labor constraints in pesticide use, thus increasing the motivation of farmers to adopt biological pesticides [28]. There is a significant negative effect of farmland size on farmers’ biological pesticide adoption behavior. Large-scale farmers have a higher level of knowledge in field pest and disease control and can effectively control pests and diseases; thus, they are less likely to adopt biological pesticides. However, smallholder farmers have a low knowledge level of field pests and lack a certain level of technical skills and experience to operate. Thus, they are more likely to use low-toxicity and high-efficiency biological pesticides for pest control [48].

Internet use has a significant positive impact on farmers’ biological pesticide adoption behavior. The Internet and social media are the main channels for farmers to obtain and learn information about biotechnology. And farmers can learn about biological pesticides and green production through the Internet, which enhances their adoption probability of biological pesticides [19]. Technical training has a significant positive effect on farmers’ biological pesticide adoption behavior. Technical training can increase the farmers’ human capital and improve their operational level in using biological pesticides, thus reducing the risk posed by biological pesticide use [3,46].

4.2.2. Endogeneity Test

This paper estimates the propensity score of service need-supply fit with the help of the logit model (See

Table A2. Impact analysis of the need-supply fit of agricultural extension services.

	Coefficient	Standard Error
Age	−0.001	0.010
Gender	−0.137	0.198
Health	−0.407 **	0.159
Education level	0.010	0.023
Part-off work	−0.039 *	0.023
Income	0.0005	0.003
Family size	−0.032	0.054
Farm size	−0.046	0.072
Quality of farmland	0.249 **	0.100
Distance to market	−0.011	0.013
Internet use	−0.294 *	0.170
Technology training	0.365 **	0.184
Jing men	2.670 ***	0.350
Jing zhou	0.954 ***	0.232
Xiang yang	−0.133	0.202
Huang gang	−0.292	0.215
Constant	0.484	0.744
Pseudo R2	0.137 ***
LR Chi2 (16)	193.98

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

). It was found that health, part-time work, and internet use had a significant negative effect on the need-supply fit of agricultural extension services. In contrast, the quality of farmland and technical training had a significant positive effect on the need-supply fit of agricultural extension services. Secondly, this study performs a balancing test on the samples using 1-nearest neighbor matching, 4-nearest neighbor matching, kernel matching, and radius matching (See

Table A3. Balance test.

Sample		Pseudo R2	LR Chi2	MeanBias	MedBias
	Unmatched	0.376	522.620 ***	33.000	34.900
1-Nearest neighbor matching	Matched	0.006	11.100	3.100	3.000
4-Nearest neighbor matching	Matched	0.004	8.350	3.000	2.700
Kernel matching	Matched	0.003	6.060	2.100	1.900
Radius matching	Matched	0.003	5.280	2.100	1.800

Note: *** indicate significance at the 1% significance levels.

). It was found that the pseudo-R² was 0.376 before matching and decreased to 0.003–0.006 after matching. And the p-value of the likelihood ratio test was significant at a 1% statistical level before matching, while none of the p-values were significant after matching. Meanwhile, MeanBias and MedBias were both below 10% after matching. This indicates that the sample passed the balance test, and the quality of matching was better.

Table 6. Endogeneity test results.

Matching Method	Propensity Score Results
	1-Nearest Neighbor Matching	4-Nearest Neighbor Matching	Kernel Matching	Radius Matching
Average treatment effect for the treated (ATT)	0.150 *** (0.045)	0.164 *** (0.038)	0.172 *** (0.036)	0.164 *** (0.037)
Control variable	Controlled	Controlled	Controlled	Controlled
Observations in the treatment group	732	732	732	732
Observations in the control group	428	428	428	428

Note: *** indicate significance at the 1% significance levels. Standard errors are in parentheses.

reports the results of the endogeneity test of service need-supply fit on farmers’ biological pesticide adoption behavior. It was found that the average treatment effect (ATT) of service need-supply fit on the adoption of biological pesticides by farmers was categorized as 0.150, 0.164, 0.172, and 0.164. This result indicates that service need-supply fit significantly contributes to the adoption of biological pesticides by farmers. The probability of adoption of biological pesticides by farmers increased by 26.29% for each unit increase in service need-supply fit. This result again verifies the robustness of the above results, and Hypothesis 1 is again validated.

4.2.3. Robustness Test

To test the robustness of the above results, this study performs robustness tests by using the replacement variable method, replacing the estimated and adjusting the sample size model (see

Table 7. Robustness test results.

Variable	Model 4		Model 5		Model 6
	Coefficient	Standard Error	Coefficient	Standard Error	Coefficient	Standard Error
Service need-supply fit	0.146 ***	0.034			1.381 ***	0.201
Number of service need-supply imbalance			−0.470 **	0.070
Control variable	Controlled	Controlled	Controlled	Controlled	Controlled	Controlled
Constant	0.045	0.101	0.088	0.963	−1.312 **	0.537
Pseudo R2	0.253 ***		0.376 **		0.378 ***
LR Chi2			522.620		524.470
F-value	22.792		-		-

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

). Firstly, this study replaces the explained variable, using ”biological pesticide use intensity” instead of “farmers’ biological pesticide adoption behavior.” From Model 4 of Table 7, we can see that there is a significant positive effect of service need-supply fit on the biological pesticide use intensity by farmers. This result is consistent with the above empirical results and indicates that the above results are robust. Secondly, this study replaces the explanatory variables, using “number of need and supply imbalance” instead of ”service need-supply fit.” From Model 6 of Table 7, there is a side argument that the higher the level of service need-supply fit, the higher the motivation of farmers to adopt biological pesticides. The third one is to replace the estimation model. This study uses the probit model to estimate the impact of service need-supply fit on farmers’ biological pesticide adoption behavior. The services need-supply fit shows a significant positive effect on the use of biological pesticides by farmers, as shown in Model 6 of Table 7. The fourth relates to the method of adjusting the sample size. This study draws on Yan et al. [46] to exclude samples older than 65 years for robustness testing, as shown in

Table 8. Robustness test of adjusted samples.

Matching Method	Propensity Score Results
	1-Nearest Neighbor Matching	4-Nearest Neighbor Matching	Kernel Matching	Radius Matching
Average treatment effect for the treated (ATT)	0.143 ** (0.056)	0.192 *** (0.047)	0.203 *** (0.043)	0.178 *** (0.047)
Control variable	Controlled	Controlled	Controlled	Controlled
Observations in the treatment group	495	495	495	495
Observations in the control group	287	287	287	287

Note: **, *** indicate significance at the 5%, 1% significance levels, respectively. Standard errors are in parentheses.

. The results found that service need-supply fit had a significant positive effect on farmers’ biological pesticide adoption behavior. The estimation results of the above robustness testing strategies are all consistent with the benchmark regression results, indicating the robustness of the benchmark regression results. Meanwhile, Hypothesis 1 is tested again.

4.2.4. Agricultural Extension Service Need-Supply Fit on Farmers’ Biological Pesticides Adoption Behavior Revisited: An Analysis Based on Threshold Regression

In order to discuss whether there is a significant difference in farmers’ biological pesticide adoption behavior under different service need-supply fit conditions, drawing on the threshold regression model proposed by Hansen [49] for cross-sectional data, we test the existence of a “tipping point” in service need-supply fit for biological pesticide adoption behavior of farmers.

$$\begin{array}{cc}{y}_i={\theta }_1^{\prime }{z}_i+{\epsilon }_i,& g_i\le \gamma \\ y_i={\theta }_2^{\prime }{z}_i+{\epsilon }_i,& g_i>\gamma \end{array}$$

(6)

In Equation (6), $y_i$ and $z_i$ denote the explained and explanatory variables, respectively. $g_i$ is the threshold variable, and the sample is divided into two groups according to the corresponding threshold value $\gamma$. ${\theta }_1^{\prime }$ and ${\theta }_2^{\prime }$ denote the regression coefficients of the two sample groups, respectively. ${\epsilon }_i$ denotes the randomized intervention group. Further, we define the dummy variable $d_i(y)=\{g_i\le Y\}$, $\{\cdot \}$ is an exponential function, and we make $Z_i(Y)=Z_i{d}_i(y)$. Then, the model for the threshold regression is set as follows.

$$y_i={\theta }^{\prime }+{\delta }_n^{\prime }{z}_i(\gamma )+{\epsilon }_i$$

(7)

Finally, using OLS methods, the corresponding threshold values $\gamma$ in Equation (7) are estimated. The bootstrap p values are calculated using the bootstrap method for significance testing of the threshold regression model, as described in Hansen [49].

Table 9. Impact of service need-supply fit on farmers’ biological pesticide adoption behavior: threshold regression results.

	Service Need-Supply Fit Is Less than or Equal 0.2	Service Need-Supply Fit Is Greater than 0.2
Service need-supply fit	−0.162 (0.266)	0.293 *** (0.073)
Control variable	Controlled	Controlled
Sample	428	732
R2	0.441	0.338
LR Chi2	68.637 ***

Note: *** indicate significance at the 1% significance levels. Standard errors are in parentheses.

reports the results of the threshold regression. The results show that when the service need-supply fit is greater than 0.2, the service need-supply fit shows a significant positive effect on farmers’ biological pesticide adoption behavior. And when the service need-supply fit is less than or equal to 0.2, the promotion effect of the service need-supply fit is not significant. It can be seen that there are significant differences in the biological pesticide adoption behavior of farmers under different service need-supply fit conditions. The higher the level of service need-supply fit, the more information farmers have about biological pesticides and the higher the probability of adopting biological pesticides.

4.2.5. Analysis of the Mechanism of Agricultural Extension Service Need-Supply Fit on Farmers’ Biological Pesticide Adoption Behavior

The previous paper better explains the direct effect of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior. However, it neglected the mechanism of influence of biological pesticide use regularization and path dependence.

Table 10. Impact of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior: an exploration of mechanisms.

	Biological Pesticide Use Regularization Model 7	Biological Pesticide Model 8	Path Dependence Model 9	Biological Pesticide Model 10
Service need-supply fit	1.219 *** (0.163)	1.230 *** (0.208)	−0.491 *** (0.142)	1.143 ** (0.208)
Biological pesticide use regularization	-	0.207 * (0.111)	-	-
Path dependence	-	-	-	−0.238 ** (0.108)
Control variable	Controlled	Controlled	Controlled	Controlled
Constant	0.180 (0.466)	−1.329 (0.553)	−0.583 (0.439)	−1.066 (0.546)
R2	0.198 ***	0.385 ***	0.052 ***	0.382 ***
LR Chi2	283.439	519.581	75.67	517.920

Note: **, *** indicate significance at the 5%, 1% significance levels, respectively. Standard errors are in parentheses.

reports the regression results of service need-supply fit on farmers’ biological pesticide adoption behavior. It can be seen that service need-supply fit significantly enhances farmers’ biological pesticide regularization behavior from Model 7. Furthermore, from Model 8, it can be seen that the degree of regularization of biological pesticide use by farmers significantly improves their motivation to adopt biological pesticides. Overall, service need-supply fit significantly increases the degree of regularization of biological pesticide use by farmers and further increases the likelihood of farmers adopting biological pesticides. Based on the above discussion, Hypothesis 2 was tested.

As shown in Model 9, service need-supply fit inhibits farmers’ path dependence on biological pesticide use. And from Model 10, farmers’ path dependence weakens their motivation to adopt biological pesticides. It can be seen that the service need-supply fit significantly weakens the path dependence of biological pesticide use by farmers, thus contributing to the likelihood of farmers adopting biological pesticides. Given the above discussion, Hypothesis 3 was tested.

5. Conclusions

Based on survey data from 1160 rice farmers in Hubei Province, China, this study uses the need-supply fit model and logit method to discuss the effect of the need-supply fit of agricultural extension services on farmers’ biological pesticide adoption behavior. Several conclusions were obtained from this study. First, the general level of need-supply fit of agricultural extension services in the study area was 36.97%. The levels of need-supply fit for the five agricultural extension services are, in descending order: pesticide dosage guidance service, pest and disease warning service, biological pesticide service, marketing service, and pesticide residue testing service. Second, health, part-off work, quality of farmland, and technical training have significant effects on the need-supply fit of agricultural extension services. Third, the agricultural extension service need-supply fit showed a significant positive correlation with farmers’ biological pesticide adoption behavior. This indicates that the higher the level of need-supply fit of agricultural extension services, the higher the motivation of farmers to adopt biological pesticides. And there were significant differences in farmers’ biological pesticide adoption behavior under different service need-supply fit conditions. When the service need-supply fit is greater than 0.2, the service need-supply fit promotes farmers’ biological pesticide adoption behavior more significantly. Fourth, the agricultural extension services need-supply fit acts on farmers’ biological pesticide adoption behavior by influencing their biological pesticide use regularization and path dependence.

Based on the above discussion, the policy suggestions of this study are as follows. First, it is recommended that agricultural bureaus, agricultural technology extension centers, agricultural stores, and other subjects should focus on promoting and applying pest and disease warning services and biological pesticide services to prevent and effectively control pests and diseases in rice fields. Secondly, during the agricultural extension process, the threshold for farmers to participate in technical training should be reduced to form a more comprehensive farmer training system. This will improve the service need-supply fit for farmers and enhance the enthusiasm of farmers to adopt biological pesticide technology. Thirdly, agricultural extension subjects can increase the applicability of technology through field guidance and technical practice to regulate farmers’ usage behavior and reduce the risk of applying new technologies. Also, they should take the initiative to master the agricultural information and services that farmers demand to alleviate the path dependence of farmers, thereby promoting a better fit between the need-supply of agricultural extension services and improving the effectiveness of agricultural extension services.

Our analysis is a cross-section of 1160 rice farmers in Hubei Province, China. Over time, there is an increasing match between the type of agricultural extension services and the content of farmers’ needs. Due to data limitations, we cannot track the dynamics of the impact of agricultural extension service need-supply fit on farmers’ biological pesticide adoption behavior. However, if applicable panel data become available in the future, we believe this will further enhance the effectiveness of our agricultural extension services. In addition, future research could explore the degree of match between service supply and farmers’ needs for agricultural mechanization services, pest control services, and other services to uncover the key factors behind the slower development process of agricultural socialization services.