Environmental Research 242 (2024) 117766

Contents lists available at ScienceDirect

Environmental Research
journal homepage: www.elsevier.com/locate/envres

How does full-cost insurance for wheat affect pesticide use? From the
perspective of the differentiation of farmers’ production scale
Yinhao Wu a, Xiangdong Duan b, Ruifeng Liu a, Hengyun Ma a, Yongmin Zhang b, *
a
College of Economics and Management, Henan Agricultural University, No.15, Longzi Lake College Park, Zhengzhou Eastern New District, Zhengzhou, 450046, China
b
Archives Center, Henan Agricultural University, No.15, Longzi Lake College Park, Zhengzhou Eastern New District, Zhengzhou, 450046, China

A R T I C L E I N F O A B S T R A C T

Keywords: Theoretically, agricultural insurance influences farmers’ use of pesticides by changing the expected income of
Full-cost insurance for wheat agricultural production. Full-cost insurance, with high guarantee and high compensation characteristics, may
Pesticide use significantly affect farmers’ pesticide use. First, this paper constructs a production function to characterize and
Large-scale farmers
compare the marginal incomes of insured and uninsured farmers under risk uncertainty and analyses how
Ordinary farmers
Simultaneous equation model
insured farmers can increase marginal income by increasing or reducing factor inputs. Considering scale dif­
ferentiation, it discusses pesticide use strategies different types of farmers may adopt to maximize household
utility. Second, using survey data of the pilot counties of full-cost insurance for wheat in Henan Province, China,
the simultaneous equation model is used for empirical testing. The results reveal the following: (i) Farmers’
insurance participation and pesticide application behaviour are not mutually independent. (ii) For the whole
sample, full-cost insurance for wheat has a significant pesticide reduction effect. (iii) However, considering scale
differentiation, pesticide application decreases significantly among insured ordinary farmers but does not change
significantly among insured large-scale farmers. Third, policy measures are proposed to activate the green
development function of agricultural insurance.

1. Introduction prevention and control measures meant to cope with this problem is
controlling the amount of pesticide inputs from the source (Ma et al.,
Pesticides are an important means of agricultural production and are 2021).
crucial to resisting diseases, pests and weeds; ensuring food production; Farmers represent the basic unit of agricultural production, and
and stabilizing farmers’ income (FAO, 2002). However, excessive use of effectively guiding their pesticide application behaviour is the key to
pesticides can lessen the number of beneficial bacterial communities in achieving the "source control of pesticide inputs". The influencing fac­
soil, trigger soil hormone organic pollution, and spread pollution to the tors of farmers’ pesticide input behaviour are diversified, but risk
entire ecological environment, such as the air, soil, and water, through avoidance and income enhancement are the dominant factors (Feng
the biological life cycle (FAO, 2002; Liu et al., 2015). Moreover, it can et al., 2021; Cole et al., 2017; Wu et al., 2018). Similarly, agricultural
increase the production costs of farmers, threaten their health, and have insurance, as a "green box policy" of government support for agriculture
a serious adverse impact on the quality and safety of agricultural allowed by WTO rules, is also expected to disperse agricultural risks and
products and international competition (Huang et al., 2003; Garming stabilize farmers’ income. Moreover, as one of the three pillars of
and Waibel, 2009; Gomes et al., 2020). The multiple negative environ­ modern agricultural development in the context of the market economy,
mental externalities caused by the excessive use of pesticides have agricultural insurance can also change income expectations by
become the primary issue restricting the sustainable development of spreading agricultural risks and affecting farmers’ input behaviour in
agriculture in China (Feng et al., 2021; Pan et al., 2021). Especially as production factors (Goodwin, 2001; Zhong et al., 2007). Put another
China enters a new stage of consolidating poverty alleviation achieve­ way, the promotion and popularization of agricultural insurance applied
ments and effectively connecting efforts with rural revitalization, it is as a tool to resolve the risk of agricultural production may change the
urgent to effectively promote pesticide reduction. One of the main patterns of farmers’ chemical usage (Quiggin et al., 1993; Goodwin

* Corresponding author.
E-mail addresses: yhwu@henau.edu.cn (Y. Wu), dxd0105@henau.edu.cn (X. Duan), ruifeng076@henau.edu.cn (R. Liu), h.y.ma@henau.edu.cn (H. Ma), dasj@
henau.edu.cn (Y. Zhang).

https://doi.org/10.1016/j.envres.2023.117766
Received 11 October 2023; Received in revised form 12 November 2023; Accepted 22 November 2023
Available online 27 November 2023
0013-9351/© 2023 Elsevier Inc. All rights reserved.
Y. Wu et al. Environmental Research 242 (2024) 117766

et al., 2004; Möhring et al., 2020). 2. Literature review

Full-cost insurance for wheat, characterized by a high guarantee and
high compensation, has a greater and more significant impact on the Scholars began to explore the impact of agricultural insurance on
production behaviour of farmers (Roberts et al., 2006; Carter et al., farmers’ production behaviour in the 1970s. Many studies have found
2016; Mao et al., 2023; 2023, Xu and Zhang). Full-cost insurance for that in the insurance market, including agricultural insurance, there is
wheat is quasi-income insurance (Jiang and Li, 2021; Claassen et al., information asymmetry between the insured and the insurer, which
2017) and can cover all the costs of wheat production, including direct induces the insured to make bold or conservative decisions, resulting in
physical and chemical costs, land costs, and labour costs. The level of moral hazard and distorting the optimal decision of the actor due to
risk protection provided is high,1 making it easier to reach the critical insurance (Knight and Coble, 1997; Goodwin and Smith, 2013). In other
level at which farmers can adjust production decisions, and the resulting words, the academic community generally agrees that agricultural in­
environmental externalities adjustments can be easily observed (2023, surance changes the production behaviour of insured farmers. However,
Xu and Zhang). From 2018 to 2020, the central government began the findings of empirical research do not yield a consensus about the
piloting the implementation of this type of insurance in Henan and effect.
Shandong Provinces, China. During this period, the full-cost insurance Empirical studies on the effect of agricultural insurance on farmers’
for wheat in Henan Province developed rapidly, playing an important pesticide use have been well documented, but the conclusions are
role in ensuring the income of food and agriculture. Subsequently, the mixed. One view is that agricultural insurance reduces pesticide inputs.
central government proposed further expanding the pilot scope and Quiggin (1992) classified agrochemicals for the first time into two
requested full coverage of full-cost insurance for wheat in major groups, namely, the increased risk type and the reduced risk type. Pes­
grain-producing counties by 2022. As the core tool of China’s agricul­ ticides are a typical reduced risk type factor (Just and Pope, 1979, 2003;
tural support and protection system, policies regarding agricultural in­ Freudenreich and Muβhoff, 2018). Under the influence of agricultural
surance should ensure both food security and green agricultural insurance, farmers tend to reduce their input of reduced risk types of
development. agricultural chemicals (Chambers, 1989; Horowitz and Lichtenberg,
As a key insurance variety to be supported and promoted by China’s 1993; Roll, 2019). Therefore, insured farmers tend to adopt a conser­
future policies, can full-cost insurance for wheat change farmers’ vative strategy that involves reducing their pesticide input (Feng et al.,
pesticide behaviour? If so, how? It is very important to deeply analyse 2021; Chakir and Hardelin, 2014). For example, traditional crop yield
this issue. This paper conducts an exploratory analysis. Unfortunately, insurance with low guarantee and low compensation significantly re­
due to the late implementation of full-cost crop insurance for crops in duces pesticide application by Chinese farmers (Mao et al., 2023). Zhong
China, the relevant practical experiences and data are insufficient (Jiang et al. (2007) found that farmers purchasing crop insurance pay
and Li, 2021), such that relevant research has not yet been conducted. In approximately 5 yuan per mu less for pesticides, approximately 19.36%
addition, coupled with the high heterogeneity of Chinese farmers (Liu, of the average, in cotton production in the Manas River Basin of Xin­
2022), it is necessary for us to comprehensively explore the impact of jiang, China. Li et al. (2022) argued that in rice production in China,
full-cost insurance for wheat on farmers’ pesticide application behav­ farmers who purchase crop insurance use 33.30% less pesticides than
iour from the perspective of farmers’ scale differentiation. farmers who do not.
This paper analyses how insured farmers increase or decrease their In contrast, the other view is that agricultural insurance does not
pesticide input to improve marginal returns by developing a production significantly decrease pesticide input but may even increase it. Under
function that can characterize and compare the marginal returns of the agricultural insurance system arrangement of low guarantee and low
insured and uninsured farmers under uncertain risks. Then, in accor­ compensation, the pesticide application behaviour of insured farmers
dance with the production scale differentiation of farmers, the effects of does not necessarily change significantly (Mishra et al., 2005). Weber
participation in full-cost insurance on the pesticide application behav­ et al. (2016) quantitatively evaluated the policy effect of expanding
iour of different types of farmers are studied, and hypotheses are pro­ agricultural insurance coverage via the U.S. Farm Bill 2014 and found
posed. Subsequently, based on the research data obtained from the first that crop insurance had little impact on pesticide use among U.S.
batch of large grain-producing counties with full-cost insurance for farmers. Moreover, Horowitz and Lichtenberg (1993) even found that in
wheat in Henan Province, a simultaneous equation model that can the Midwest U.S., corn growers who purchase crop insurance spend
address multiple endogeneity problems is used for empirical testing. more on pesticides per acre. Chang and Mishra (2012) also confirmed a
Finally, combined with the new concept of ecological priority and green positive effect of crop insurance on expenditures on chemicals per acre
development, policies and measures are proposed to effectively stimu­ in the United States. He et al. (2020) concluded that cost-of-production
late the green development of agricultural insurance. insurance significantly increased the total expenditure of Philippine
The structure of this paper is as follows. In Sec. 2, we present a review corn farmers on chemicals, such as pesticides, and the same was true in
of the literature on the impacts of agricultural insurance. In Sec. 3, the Bangladesh (Hill et al., 2019).
theoretical analysis, we construct a production function and a household In summary, although the academic community generally agrees
utility function to theoretically analyse the changes in the pesticide that agricultural insurance can spread risks, change income expecta­
application behaviour of farmers of different farming scales after tions, and then affect farmers’ pesticide input behaviour, empirical
participating in full-cost insurance and propose hypotheses. The studies find mixed effects. The reasons for this are as follows: (1) It is
empirical models and estimation results are presented and discussed in misleading to compare the research conclusions in a general manner
Sec. 4. We conclude in Sec. 5 by summarizing major findings and without distinguishing the types of insurance. Different types of insur­
highlighting policy implications. ance have different effects on farmers’ agrochemical use (Walters et al.,
2012; Carter et al., 2016). The higher the insurance guarantee is, the
more likely it is to distort farmers’ pesticide application behaviour
(Roberts et al., 2006). As an example, compared to traditional agricul­
1
tural insurance, full-cost insurance has a greater impact on farmers’
Taking wheat as an example, in China, the coverage amount of traditional
production behaviour. It is necessary to distinguish between these types
yield insurance is only 447 yuan/mu, with an average premium of 27 yuan/mu.
of insurance. Unfortunately, existing research focuses on traditional
In contrast, the coverage amount of full-cost insurance for wheat reaches 1000
yuan/mu, with an average premium of 45 yuan/mu. This insurance is a policy insurance types with low guarantee and low compensation, but studies
based agricultural insurance purchased by farmers voluntarily. The central and analysing how full-cost insurance affects farmers’ pesticide use are rare
provincial governments provide 70% subsidy for insurance premiums, and in China. (2) Many relevant studies overlook the impact of farmer het­
farmers pay 30% on their own. erogeneity. However, differences in farmers’ production scale can alter

2
Y. Wu et al. Environmental Research 242 (2024) 117766

the impact of crop insurance on their pesticide application behaviour. production of uninsured farmers is E(R):
(3) The empirical methods used in the literature often ignore or fail to
E(R)= ∅ pf(X, D1 ) + (1 − ∅ )pf(X, D2 ) − vX (1)
effectively address model endogeneity issues (Zhong et al., 2007; Zhang
et al., 2018). However, due to the bidirectional causal relationship be­ where p, Ф, and ν represent the price of a certain agricultural product,
tween farmers’ insurance decisions and agrochemical use behaviour, the probability of no disaster occurring, and the price of production
ignoring this will lead to bias in the estimation results (Zhong et al., factors, respectively. The marginal income of the production factor input
2007; Zhang et al., 2018). of uninsured farmers is:
The marginal contribution of this article lies in the following: First,
research object innovation. We focus on the pesticide application R′= ∅ pf′(X, D1 ) + (1 − ∅ )pf′(X, D2 )− v (2)
behaviour of farmers participating in full-cost insurance for wheat and
On the other hand, the expected income function of the agricultural
supplement the existing research on the impact of high guarantee and
production of insured farmers is E(R*):
high compensation insurance types on farmers’ agrochemical use
behaviour. Second, research perspective innovation. Based on the E(R∗ )= ∅ pf(X, D1 ) + (1 − ∅ )pf(X, D2 ) + (1 − ∅ )M0 ϑ(X, D2 ) − vX− c
perspective of farmer differentiation, this paper examines the impact of (3)
full-cost insurance for wheat on the pesticide use behaviour of different
types of farmers. It provides a basis for improving insurance clause where ϑ(X, D2 ), M0 , c represent the loss rate, the insurance amount of
design and government management. Third, research method innova­ different growth periods of food crops, and the premium. The loss rate ϑ
tion. The study uses a simultaneous equation model (SEM) that can is a function of the actual yield of crops at the time of a disaster, g((Y −
effectively address multiple endogeneity problems for empirical testing, f(X,D2 )) /Y), and is determined by comparing the actual yield after the
offering a new research method for accurately evaluating the effec­ disaster with the local average output in recent years. Then, the mar­
tiveness of agricultural insurance policy. ginal income of the production factor input of the insured farmers is:

3. Analysis and research hypotheses R∗′= ∅ pf′(X, D1 ) + (1 − ∅ )pf′(X, D2 ) + (1 − ∅ )M0 ϑ′(X, D2 )− v (4)
Finally, the marginal benefits of the production factors of the insured
Moral hazard and adverse selection caused by information asym­
and uninsured farmers are compared: ΔR′ as follows:
metry are the theoretical basis for studying the effect of insurance on the
behaviour of the insured (Arrow, 1963; Smith and Goodwin, 1996). ΔR′ = R∗′ − R′ = (1 − ∅ )M0 ϑ′(X, D2 ) (5)
Moral hazard exists objectively in the insurance market (Miranowski,
1974) and can induce participants to adopt aggressive or conservative From formula (5), it is not difficult to conclude that farmers’
behaviour (Knight and Coble, 1997; Goodwin and Smith, 2013), which participation in agricultural insurance affects the marginal income of
increase with the improvement in the insurance guarantee level (Rob­ their production factors and then drives them to change the factor input.
erts et al., 2006; 2023, Xu and Zhang). Due to the uncertainty of the That is, if ΔR′> 0, farmers will increase factor input; otherwise, they will
impact of moral hazard on farmers’ production behaviour, farmers reduce factor input. Furthermore, combined with Quiggin’s (1992)
participating in agricultural insurance may make different input de­ classification of production factor types, rational insured farmers opt to
cisions for different agricultural chemicals. increase the input of strongly increased risk type production factors,
According to the basic economic assumption of the "rational man", reduce the input of reduced risk type factors, and appropriately increase
whether insured farmers are willing to increase the input of a certain or maintain the input of weakly increased risk type factors. As pesticide
production factor depends on whether increasing this factor increases is a reduced risk type factor (Just and Pope, 2003; Freudenreich and
their marginal output. On this basis, referring to the ideas and methods Muβhoff, 2018), this paper proposes H1:
of Quiggin (1992), Horowitz and Lichtenberg (1993), and Yan and Keith H1. Insured farmers tend to reduce their pesticide input.
(2009), this paper attempts to construct a production function that can However, this conclusion is based on the assumption that farmers are
compare the marginal returns of factor inputs between insured and homogenous. In fact, Chinese farmers are highly differentiated (Liu,
uninsured farmers under uncertain risk conditions. 2022), such that different types of insured farmers present different
First, the production function of the farmer is set as Yi = f(X, Di ), i = pesticide application behaviour. From the perspective of scale differ­
1, 2, where X represents the matrix of production factors. For conve­ entiation, this paper categorizes farmers into large-scale farmers and
nience of expression, it is assumed that farmers use only one factor of ordinary farmers according to their planting scale (Table 1). These two
production (in fact, when multiple inputs are needed in agricultural types of farmers have different goals for agricultural production. The
production, X can be regarded as a combination of these factors). Di former intend to maximize agricultural returns, while the latter aim to
represents two uncertain production environments confronted by rationally allocate household labour resources and achieve food
farmers; namely, D1 indicates that there is no disaster in the year, and D2 self-sufficiency (MARAPRC, 2013).
indicates that there are disasters in that year. Usually, in the absence of Based on this, we further construct the total effect function of
disasters, the marginal output of production factors is positive, i.e., different types of farmers. According to the principle of optimal allo­
f′(X, D1 ) > 0. cation of asset portfolios, the total utility of the economic subject is a
Second, based on the classification of agricultural factors by Quiggin function of the expected return and return fluctuation. That is,
(1992) and Quiggin et al. (1993), if f′(X, D1 ) ≤ f′(X, D2 ), then the pro­
duction factor is defined as belonging to the reduced risk type; other­ U = U[E(π), Var(π)] (6)
wise, if f′(X,D1 ) ≥ f′(X,D2 ), the production factor is defined as belonging where U represents the total utility of farmers, E(π) represents the ex­
to the increased risk type. Particularly, the increased risk types of pro­ pected returns of farmers, and Var(π) represents the variance of farmers’
duction factors can be further divided into the weakly increased risk returns. Obviously, the utility function U satisfies:
type, i.e., f′(X, D2 ) ≥ 0, and the strongly increased risk type, i.e.,
∂U ∂U
f′(X, D2 ) < 0. In the context of the pursuit of profit maximization, the > 0 and <0 (7)
∂E(π) ∂Var(π)
expected return function and factor marginal return function can be
obtained for uninsured farmers and insured farmers who are engaged in
That is, the total utility of the household is an increasing function of the
agricultural production.
expected return of the household and a decreasing function of the
On the one hand, the expected income function of the agricultural

3
Y. Wu et al. Environmental Research 242 (2024) 117766

Table 1 Therefore, to maximize their incomes, the large-scale farmers who

Variable definitions and descriptive statistics. participate in full-cost insurance do not have the motivation to actively
Variable type and variable definitions Variable Mean Standard reduce ex .
Code Deviation

Dependent Variables (2) The expected income function of ordinary farmers who partici­
Frequency of pesticide use per mu of pest1 3.516 1.822 pate in full-cost insurance is:
wheat planting by the full sample of
farmers < listaend > E2 (π)= θ[pM(ex1 )+ω(ex2 )− C(ex )
Frequency of pesticide use per mu of pest11 3.794 1.659
− τ0 ] + (1 − θ)[Y∗ +ω(ex2 )− C(ex ) − τ0 ]
wheat planting by large-scale farmers
Frequency of pesticide use per mu of pest12 3.450 1.901
wheat planting by ordinary farmers = θpM(ex1 ) + (1 − θ)Y∗ +ω(ex2 )− C(ex ) − τ0 (11)
Independent Variables
Participation in full-cost insurance for Dinsu 0.391 0.489 where ex = ex1 +ex2 ; ex1 , ex2 represent the efforts of ordinary farmers in
wheat? (Yes = 1; otherwise = 0) agricultural and nonagricultural work, respectively; ω(ex2 ) represents
Large-scale farmer?a (Yes = 1; otherwise Dscal 0.191 0.393
the income from off-farm or part-time work for the ordinary farmers.
= 0)
Control Variables Obviously, ω′(ex2 )>0.
Has the farmer previously received Dpay 0.632 0.483 According to the Lagrange extreme value theorem, the conditions for
agricultural insurance compensation? the maximum expected income of insured ordinary farmers are:
(Yes = 1; otherwise = 0)
Have past natural disasters have had a Ddisa 0.715 0.452 ∂E2 (π)
significant impact on the farmer’s = θpM′(ex1 ) − ω′(ex2 ) − C′(ex ) = 0, that is,θpM′(ex1 ) = ω′(ex2 ) + C′(ex )
∂ex1
wheat production? (yes = 1; otherwise
= 0) (12)
Does the farmer understand full-cost Dcong 0.527 0.362
insurance for wheat? (Yes = 1;
Therefore, to maximize income, ordinary farmers who participate in
otherwise = 0) full-cost insurance will invest less effort (ex1 ) in agricultural production
Head of household age (years) age 47.429 15.837 (and more effort in nonagricultural production) to improve their own
Has the head of household received a high Deduc 0.303 0.460 agricultural marginal income, which satisfies equation (12). That is,
school education (including vocational
ordinary farmers who participate in full-cost insurance have the moti­
school) or above? (yes = 1; otherwise =
0) vation to reduce their agricultural efforts.
Has the head of household received Dtrai 0.248 0.432
agricultural technical training, such as (3) Compare Var(π) and ex for the two types of farmers. The sensi­
scientific pesticide use methods? (yes = tivity of large-scale farmers to agricultural income risks is much
1; otherwise = 0)
greater than that of ordinary farmers. Therefore, even if insured,
a
Referring to the definition of scale operation by the Food and Agriculture the former will not reduce ex to reduce Var(π), but the latter may
Organization of the United Nations, with a threshold of 2 ha, planting more than do the opposite.
2 ha is considered scale management.
Based on the above analysis, under the objective function MaxU =
variance of the return. U[E(π),Var(π)], the level of agricultural production efforts of large-scale
Therefore, the objective function of farmers to change their pro­ farmers will not decrease after participating in insurance. However,
duction behaviour is: ordinary farmers have the motivation to shift more efforts to nonagri­
MaxU= U[E(π), Var(π)] (8) cultural production to improve their overall utility.
Accordingly, this paper proposes H11 and H12:
Next, consider how different types of farmers can achieve utility
maximization by changing their production behaviour. H11. Insured large-scale farmers tend not to reduce their pesticide
input.
(1) Expected return function of large-scale farmers who participate in H12. Insured ordinary farmers tend to reduce their pesticide input.
full-cost insurance: E1 (π). In particular, compared with traditional yield insurance, which is
E1 (π)= θ[pM(ex )− C(ex ) − τ0 ] + (1 − θ)[Y∗ − C(ex ) − τ0 ] characterized by low guarantee and low compensation, full-cost insur­
ance for wheat is a type of insurance with high guarantee and high
= θpM(ex ) + (1 − θ)Y∗ − C(ex ) − τ0 (9) compensation, which may have more widely varying effects on the
production behaviour of different types of insured farmers.
where ex represents the various inputs of farmers in agricultural pro­
duction (such as materialized costs, time management costs), and for 4. Research design
ease of expression, we define ex as the effort level of farmers.
M(ex ) and C(ex ) represent the yield and production cost of agricultural 4.1. Data source
products, respectively, under the condition that the level of effort of
farmers is ex . M(ex ) and C(ex ) are both increasing functions of ex . p andτ0 Henan is the largest wheat-growing province in China, accounting
represent the price of agricultural products and the full-cost insurance for more than 28% of the country’s total wheat output. It is also one of
premium. Θ represents the probability of no insurance compensation the only two provinces piloting full-cost insurance for wheat in China.
occurring. Y∗ represents the total income received by farmers after During the pilot period from 2018 to 2021, full-cost insurance for wheat
receiving insurance compensation. in the Henan pilot area developed rapidly, the insured area increased by
According to the Lagrange extreme value theorem, the conditions for more than 113%, and the claim amount increased by 189%, which
the maximum expected income of insured large-scale farmers are as played an important role in protecting the income of farmers in Henan
follows: Province, China. Taking Henan as an example for studying the influence
of full-cost insurance for wheat on farmers’ pesticide application
∂E1 (π)
= θpM′(ex ) − C′(ex ) = 0, namely,θpM′(ex ) = C′(ex ) (10) behaviour has both natural advantages and is very representative.
∂ex
Therefore, from June to August 2021, this research group adopted a

4
Y. Wu et al. Environmental Research 242 (2024) 117766

stratified sampling method to investigate the six major grain counties, were described and statistically analysed (Table 1).
namely, Yanling County, Xiangcheng County, Ruzhou County, Xiuwu Table 1 confirms that the samples selected in this paper basically
County, Wuzhi County and Qi County, that took the lead in imple­ conform to the development status of wheat cultivation in Henan
menting full-cost insurance for wheat. A total of 500 questionnaires Province. Overall, approximately 39% of the region’s farmers have
were distributed, and 457 questionnaires were recovered. Unreliable purchased full-cost insurance for wheat. The number of ordinary farmers
samples with missing data or obvious abnormal values were excluded, accounts for 81% of the full sample, while large-scale farmers account
393 valid questionnaires were finally obtained, and the effective rate of for only 19%. Wheat production is sensitive to the impact of natural
the samples was 86%. disasters, and the majority of farmers purchase agricultural insurance
(mainly traditional yield insurance). However, the awareness of full-cost
4.2. Variable selection insurance for wheat is low, as are the cultural level and the opportunity
to receive agricultural technical training.
The relevant variables are set around the core issue of the
research—the question of whether and how farmers’ pesticide input 5. Estimated results and analysis
behaviour changes after the purchase of full-cost insurance for wheat­
—and are presented below. 5.1. Endogeneity test and estimation method
Dependent variable: The frequency of pesticide use by wheat
growers is set as the dependent variable to characterize their pesticide Before using the simultaneous equation model for empirical analysis,
application behaviour. The reason is that there are significant differ­ it is necessary to test the endogeneity of variables to ensure the simul­
ences in the composition and content of different pesticides, and it is taneity of equations. The Hausman test results show that the simulta­
difficult to effectively distinguish the differences in pesticide use neous equation model passes the endogeneity test, indicating that
behaviour among farmers by using the average amount of pesticides per equations (13) and (14) have simultaneous conditions. On this basis, this
mu. article adopts the three-stage least squares method (3SLS) for model
Independent variables: The core explanatory variables of this paper regression. As a complete information estimation method, 3SLS can
are set as whether farmers have purchased full-cost insurance for wheat simultaneously solve problems such as simultaneous biases within
and the production scale of farmers. The characteristic variables of scale equations and correlations between equations, ensuring more effective
differentiation include large-scale farmers and ordinary farmers. results.
Control variables: In addition to the above factors, other factors
affect farmers’ pesticide application behaviour. In reference to relevant
studies (Zhong et al., 2007; Cole et al., 2017; Zhang et al., 2020; Hui 5.2. Analysis of empirical results
et al., 2023), the following control variables are included in the model:
whether the farmer previously received agricultural insurance The 3SLS estimation method is used for simultaneous equation
compensation, the impact of the past natural disasters on household regression analysis of equations (13) and (14), and the estimation results
wheat production, familiarity with full-cost insurance for wheat, age of are shown in Table 2 and Table 3
householder, cultural level, prior agricultural technical training, such as
scientific fertilization methods, etc. In addition, there may be regional 5.2.1. Changes in pesticide use behaviour of farmers in the whole sample
differences in pesticide use, so regional control variables are added to
the model. (1) Full-cost insurance for wheat has induced moral hazard and
changed farmers’ pesticide application behaviour. After partici­
pating in full-cost insurance for wheat, farmers’ pesticide use
4.3. Model construction
significantly decreased. Column 2 of Table 2 shows that the fre­
quency of pesticide use by insured farmers is significantly less
First, the econometric regression equation indicating how farmers’
than that of uninsured farmers at the statistical level of 5%. With
insurance participation affects their pesticide input behaviour is con­
a one standard-deviation increase in the full-cost insurance for
structed as follows:
wheat coverage rate, the frequency of pesticide use reduces by
pest= f(insu,scal, Λ) (13)
Table 2
where pest represents the intensity of the pesticide use of farmers, insu Effect of full-cost insurance for wheat on farmers’ pesticide input.
represents whether the farmers are insured, scal represents the scale
Variables Full sample of farmers
differentiation characteristics of farmers, and Λ is the control variable.
Second, an effective estimation method is chosen. There may be pest1 Dinsu

serious endogeneity problems between the dependent and independent 1 2 3

variables in equation (13). In fact, there is a logical two-way causal Dinsu − 0.235** (0.116) –
pest1 − 0.0917* (0.052)
relationship between farmers’ participation in agricultural insurance –
pest11 – –
and their pesticide input decisions (Mishra and Goodwin, 2006; Zhong pest12 – –
et al., 2007). Using the ordinary least squares method (OLS) may lead to Dscal 0.031 (0.437) 0.462*** (0.076)
biased and inconsistent estimation results. We further construct the Ddisa 0.694*** (0.128) 0.570*** (0.151)
econometric regression equation of pesticide application behaviour Dpay – 0.688*** (0.224)
Dcong 0.493*** (0.106)
affecting farmers’ insurance participation and use the simultaneous
–
age 0.000 (0.206) − 0.000 (0.083)
equation model to empirically test equations (13) and (14). Deduc − 0.125*** (0.045) 0.431** (0.189)
Dtrai − 0.383*** (0.079) 0.046* (0.020)
insu= f(pest,scal, Θ) (14) Regional Various control control
Cons. 171.204*** 103.115***
where Θ represents the control variable. R2 0.358 0.602
Obs. 393 393
4.4. Descriptive statistics Notes: Values in brackets are the standard error of the regression coefficients;
***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively
A total of 393 valid questionnaires were processed, and the variables (the same below).

5
Y. Wu et al. Environmental Research 242 (2024) 117766

Table 3 the frequency of pesticide use. Therefore, taken together, the

Effects of full-cost insurance for wheat on pesticide use by farmers with different frequency of pesticide use between the two is similar.
scales.
Variables Large-scale farmers Ordinary farmers 5.2.2. Changes in pesticide use behaviour of insured farmers with different
pest11 Dinsu pest12 Dinsu
production scales

1 2 3 4 5
(1) The total household utility of large-scale farmers is highly
Dinsu 0.017 (0.100) – − 0.291*** –
(0.085) dependent on agricultural income, and it is difficult to induce
pest1 – – – – serious moral hazard after participating in insurance. Thus, there
pest11 – − 0.058* – – is no significant change in pesticide application behaviour. Ac­
(0.035) cording to column 2 of Table 3, for large-scale farmers, whether
pest12 – – – − 0.163**
（0.074）
or not they are insured does not significantly change their
Dscal – – – – pesticide use behaviour. This confirms H11: insured large-scale
Ddisa 0.907*** 0.690*** 0.533* 0.502*** farmers tend not to reduce their pesticide input. The reason is
(0.152) (0.147) (0.314) (0.161) that the household income of large-scale farmers mainly comes
Dpay 0.718*** 0.596***
– –
from agricultural production, and their core goal in engaging in
(0.133) (0.201)
Dcong – 0.585*** – 0.378** agricultural production is to pursue the maximization of agri­
(0.101) (0.156) cultural income. Therefore, even with the purchase of full-cost
age 0.000 (0.212) − 0.000 0.000 (0.280) − 0.000 insurance for wheat, large-scale farmers remain unwilling to
(0.106) (0.234) reduce the risk-reducing factor input of pesticides in the pro­
Deduc − 0.073* 0.569*** − 0.140 0.314*
(0.041) (0.190) (0.836) (0.185)
duction process and instead use insurance more as an income
Dtrai − 0.296* 0.073** − 0.521 0.022 (0.169) "bottom cover" tool. In addition, in the process of investigation,
(0.172) (0.036) (0.491) we even found that some policyholders choose to exert the
Regional control control control control transfer payment effect of premium subsidies (Smith and Good­
Various
win, 1996) and increase pesticide use.
Cons. 153.798*** 294.275*** 215.130*** 127.563***
R2
0.353 0.551 0.302 0.476 (2) The total household utility of ordinary farmers is less dependent
Obs. 75 75 318 318 on agricultural income, and they often adopt the "one household,
two systems" production mode2 (Xu et al., 2013), which signifi­
cantly reduces their frequency of pesticide application. According
approximately 0.235 times. This confirms H1: insured farmers to column 4 of Table 3, for ordinary farmers, participating in
tend to reduce their pesticide input. The reason is that, on the full-cost insurance for wheat can significantly change their
whole, full-cost insurance for wheat induces moral hazard and pesticide use behaviour. This confirms H12: insured ordinary
changes farmers’ pesticide application behaviour. That is, after farmers tend to reduce their price input. It also confirms the
participating in insurance, farmers’ risk prevention and man­ research conclusion of Chang and Mishra (2012) that full-cost
agement of wheat production become lax. In fact, this conclusion insurance can significantly reduce the intensity of pesticide
has been confirmed in both traditional policy-based agricultural application for ordinary farmers, which is still valid in the pri­
insurance and full-cost insurance in developed countries (Chang mary market of full-cost insurance in China. The reasons for this
and Mishra, 2012). are as follows: first, the proportion of agricultural income in the
(2) The pesticide application frequency of farmers induces adverse household income of ordinary farmers is relatively low, and one
selection, which affects their insurance participation decisions. of their main goals of engaging in agricultural production is to
With the increase in pesticide use, the probability of farmers regulate the rational allocation of household labour resources.
participating in full-cost insurance for wheat decreased signifi­ Therefore, when uninsured, such farmers often use pesticides and
cantly. Column 3 of Table 2 shows that at the statistical level of other agricultural chemicals as substitutes for time and
10%, the frequency of pesticide application per acre significantly manpower input in agricultural production management (Wu
reduces the insurance participation rate of farmers. The more et al., 2018), adopting a "one shot" approach to heavily use
frequently farmers use pesticides, the less likely they are to pur­ agricultural chemicals. However, when ordinary farmers partic­
chase full-cost insurance for wheat. In fact, for every increase in ipate in insurance, they may instead view agricultural insurance
pesticide use, the insured rate of farmers decreased by approxi­ as a substitute for production management and factor inputs and
mately 9.17 percent, which confirms that the frequency of reduce inputs of agricultural chemicals, such as pesticides. Sec­
farmers’ pesticide application induces adverse selection in in­ ond, ordinary farmers’ main goal of planting wheat is to satisfy
surance participation decisions. In addition, among the control self-sufficiency. Therefore, regardless of whether they are insured
variables, only Ddisa, Deduc and Dtrai have significant effects on or not, ordinary farmers adopt green production methods to
farmers’ pesticide application behaviour. Specifically, farmers cultivate their own edible agricultural products. Third, the
with high school education and above, as well as farmers who pesticide use behaviour of ordinary farmers exhibits a relatively
have received agricultural technical training, use pesticides obvious characteristic of early empirical fixed behaviour habits
significantly less frequently. Farmers who believe that past nat­ and is easily influenced by surrounding farmers (Wu and Pretty,
ural disasters have had a significant impact on their wheat pro­ 2004), with a "herd effect". This neutralizes the impact of farmers’
duction tend to use pesticides more frequently. Notably, there is participation in insurance to some extent.
no significant difference in the frequency of pesticide use be­
tween large-scale farmers and ordinary farmers. The reason is
that although large-scale farmers are more focused on agricul­
2
tural risk management, they use pesticides more frequently. The "one household, two systems" mode of production refers to the fact that
However, at the same time, in production, large-scale farmers are most farmers apply large amounts of pesticide and fertilizer only to agricultural
products sold in the market and adopt a green production mode for their own
more willing and able to use comprehensive pesticide application
edible agricultural products. In other words, farmers generally have the agri­
methods, such as "one spray and three prevention", thus reducing
cultural production behaviour of ensuring output externally and quality
internally.

6
Y. Wu et al. Environmental Research 242 (2024) 117766

(3) Farmers’ frequency of pesticide application at different produc­ significant role in promoting pesticide reduction. Therefore,
tion scales significantly affects their insurance decision-making, further expanding the coverage of full-cost insurance for wheat
exhibiting adverse selection. As the frequency of pesticide use has positive significance for promoting the green transformation
increases, the likelihood of farmers of different production scales of agriculture in Henan and even throughout China. The key to
participating in full-cost insurance for wheat significantly de­ enabling more farmers to participate in full-cost insurance for
creases. Among them, with a one standard-deviation increase in wheat is to improve the service awareness and level of insurance
the frequency of pesticide use, the participation rate of large-scale institutions. Only when farmers receive better insurance services
farmers decreases by approximately 5.8%, and the participation and tangible compensation for agricultural production losses will
rate of ordinary farmers decreases by approximately 16.3%. This they have a stronger willingness to become insured.
further confirms that regardless of the production scale, there is (2) Insurance institutions should establish a premium subsidy
adverse selection in the insurance decisions of farmers. Among mechanism for agricultural green production. To stimulate the
the control variables, Ddisa has a significant impact on pesticide green transformation motivation of insured farmers, insurance
application behaviour among farmers of different scales but has a companies should set restrictive requirements in the clauses
greater impact on large-scale farmers, reflecting a higher sensi­ regarding full-cost insurance for wheat; that is, farmers can
tivity of large-scale farmers to agricultural production risks. Deduc receive premium financial subsidies only when they meet green
and Dtrai reduce the frequency of pesticide application for large- production standards. The study indicates that although the full-
scale farmers but do not significantly change the pesticide cost insurance for wheat promoted a reduction in pesticide use in
application behaviour of ordinary farmers. Henan wheat production overall, the main body of the reduction
concerned insured ordinary farmers, and the same effect was not
5.3. Robustness tests observed for large-scale farmers. In other words, it is the moral
hazard of full-cost insurance for wheat that currently incentivizes
The above empirical conclusions are tested for robustness by ordinary farmers to use less pesticide. However, the agricultural
replacing the dependent variables and the regression model. The fre­ green transformation mechanism utilizing "full-cost insur­
quency of pesticide application among wheat growers is compared with ance→adjusting farmers’ production behaviour→promoting
the average of all surveyed wheat farmers. If the frequency is greater pesticide reduction" has not yet been established. Therefore, it is
among the former than the latter, it is counted as 3; if it is equal, it is necessary to further improve the full-cost insurance for the wheat
counted as 2; and if it is less, it is counted as 1. Then, the endogenous system in the future and stimulate the green transformation
ordered probit model is applied for regression analysis, and the obtained motivation of all insured farmers, especially large-scale farmers,
conclusions are basically consistent with the above conclusions. Among by setting restrictive clauses regarding green production of
them, the direction and significance of the regression coefficients of the financial subsidies for premiums.
core explanatory variables remain unchanged. The significance of age (3) Farmers’ awareness of full-cost insurance for wheat should be
and Deduc among the control variables is slightly different. The robust­ improved. China has long implemented an agricultural insurance
ness test further confirms the three hypotheses proposed in this paper. system with low guarantee, low compensation, and wide
coverage. Farmers’ general awareness of agricultural insurance
6. Conclusion and recommendations types with high guarantee and high compensation is relatively
low, which greatly affects their production behaviour changes
This paper first constructs a production function that can compare after participation in insurance. Therefore, it is necessary to
the marginal income of the production factor input levels of insured and support farmers’ comprehensive and accurate understanding of
uninsured farmers under uncertain risk. Then, under the assumption of full-cost insurance for wheat through extensive publicity and
the "rational man", the means by which insured farmers can improve the specialized education to clarify the essential difference between
marginal income of production factors by increasing or decreasing factor traditional yield insurance, with low guarantee and low
inputs is analysed. Based on strict assumptions, this paper theoretically compensation, and full-cost insurance for wheat, with high
proves that, as a decreased risk type production factor, farmers tend to guarantee and high compensation. This can support the active
reduce their use of pesticides under the action of agricultural insurance. integration of more insured farmers into the green transformation
However, considering that farmers are highly differentiated in China of agriculture.
(Liu, 2022), this paper divides farmers into two types: large-scale
farmers and ordinary farmers. To maximize the total household utility, Finally, in this paper, farmers are divided into scale farmers and
different types of insured farmers make different decisions on pesticide ordinary farmers from the perspective of scale differentiation, but there
use under different agricultural production goals, which leads to the may be room for further classification. Ordinary farmers can be further
establishment of the hypotheses of this paper. Finally, based on the divided into a first type of part-time farmers based on agricultural pro­
survey data of the pilot counties of full-cost insurance for wheat in duction and a second type of part-time farmers based on nonagricultural
Henan Province, China, this article uses a simultaneous equation model production. By considering the perspective of scale differentiation and
that can simultaneously address multiple endogeneity problems for part-time differentiation, we may obtain a more accurate conclusion.
empirical testing and draws the following main conclusions. (1) This will be the direction of our next in-depth study.
Farmers’ insurance participation and pesticide application behaviour
are not mutually independent. (2) From the perspective of the entire Credit authorship contribution statement
sample of farmers, full-cost insurance for wheat has a significant pesti­
cide reduction effect. (3) However, from the perspective of scale dif­ Conceptualization, Y.W. and H.M.; methodology, Y.W. and Y.Z.;
ferentiation, the pesticide application of insured ordinary farmers formal analysis, X.D.; investigation, Y.W., X.D. and Y.Z.; data curation,
decreases significantly, but that of insured large-scale farmers does not Y.W. and X.D.; writing—original draft preparation, Y.W.; writing—re­
change significantly. view and editing, Y.Z.; language polishing, H.M. and R.L.; supervision,
The policy implications of this article are as follows: Y.Z.; funding acquisition, Y.W. All authors have read and agreed to the
published version of the manuscript.
(1) The central government should further expand the coverage of
full-cost insurance for wheat. From the entire sample of farmers
in the Henan pilot area, full-cost insurance for wheat has played a

7
Y. Wu et al. Environmental Research 242 (2024) 117766

Declaration of competing interest Knight, T.O., Coble, K.H., 1997. Survey of US multiple peril crop insurance literature
since 1980. Rev. Agric. Econ. 19 (1), 128–156.
Li, H.J., Yuan, K.H., Cao, A.D., 2022. The role of crop insurance in reducing pesticide use:
The authors declare the following financial interests/personal re­ evidence from rice farmers in China. J. Environ. Manag. 306 (1–3), 114456.
lationships which may be considered as potential competing interests: Liu, S.Y., 2022. The fundamental force determining rural changes is the differentiation
Yinhao Wu reports financial support was provided by China Post­ and intergenerational changes of farmers. Xinhua Daily, 2022-04-29 (13th edition).
(in Chinese).
doctoral Science Foundation. Yinhao Wu reports administrative support Liu, Y.B., Pan, X.B., Li, J.S., 2015. A 1961-2010 record of fertilizer use, pesticide
was provided by The National Social Science Fund of China. Yinhao Wu application and cereal yields: a review. Agron. Sustain. Dev. 35 (1), 83–93.
reports administrative support was provided by Philosophy and Social Ma, J.J., Yang, C., Cui, H.Y., Wang, X., 2021. The environmental effect and formation
mechanisms of the promotion of agricultural insurance: from the perspective of non-
Science Foundation of Henan Province. point source pollution of chemical fertilizers in China. Insur. Stud. 401 (9), 46–61 (in
Chinese).
Data availability Mao, H., Chen, S., Ying, R., Fu, Y., 2023. How crop insurance influences agrochemical
input use: evidence from cotton farmers in China. Aust. J. Agric. Resour. Econ. 67
(2), 224–244.
The authors do not have permission to share data. MARAPRC, 2013. Ministry of Agriculture and Rural Affairs of the People’s Republic of
China). Changes in the main body of grain production and management and their
impact on food security. Agric.Prod. Market 634 (28), 22–25 (in Chinese).
References Miranowski, J.A., 1974. Crop Insurance and Information Services to Control Use of
Pesticides. Office of Research and Development, U.S. Environmental Protection
Hui, M., Chen, S.J., Ying, R.Y., Fu, Y., 2023. How crop insurance influences agrochemical Agency, Washington.
input use: evidence from cotton farmers in China. Aust. J. Agric. Resour. Econ. 67 Mishra, A.K., Goodwin, B.K., 2006. Revenue insurance purchase decisions of farmers.
(1), 1–21. Appl. Econ. 38 (2), 149–159.
Arrow, K.J., 1963. Uncertainty and the welfare economics of medical care. Am. Econ. Mishra, A.K., Nimon, R.W., El-Osta, H.S., 2005. Is moral hazard good for the
Rev. 55 (1/2), 154–158. environment? Revence insurance and chemical input use. J. Environ. Manag. 74 (1),
Carter, M.R., Cheng, L., Sarris, A., 2016. Where and how index insurance can boost the 11–20.
adoption of improved agricultural technologies. J. Dev. Econ. 118 (C), 59–71. Möhring, N., Dalhaus, T., Enjolras, G., Finger, R., 2020. Crop insurance and pesticide use
Chakir, R., Hardelin, J., 2014. Crop Insurance and pesticide use in French agriculture: an in European agriculture. Agric. Syst. 184 (7), 102902.
empirical analysis. R. Agric. Environ. Stud. 95 (1), 25–50. Pan, Y., Ren, Y., Luning, P.A., 2021. Factors influencing Chinese farmers’ proper
Chambers, R.G., 1989. Insurability and moral hazard in agricultural insurance markets. pesticide application in agricultural products - a review. Food Control 122 (4),
Am. J. Agric. Econ. 71 (3), 604–616. 107788.
Chang, H.H., Mishra, A.K., 2012. Chemical usage in production agriculture: do crop Quiggin, J., 1992. Some observations on insurance, bankruptcy and input demand.
insurance and off-farm work play a part? J. Environ. Manag. 105 (4), 76–82. J. Econ. Behav. Organ. 18 (1), 101–110.
Claassen, R., Langpag, C., Wu, J.J., 2017. Impact of federal crop insurance on land use Quiggin, J., Karagiannis, G., Stanton, J., 1993. Crop insurance and crop production: an
and environmental quality. Am. J. Agric. Econ. 99 (3), 592–613. empirical study of moral hazard and adverse selection. Aust. J. Agric. Econ. 37 (2),
Cole, S., Gine, X., Vickery, J., 2017. How does risk management influence production 95–113.
decisions? Evidence from a field experiment. Rev. Financ. Stud. 30 (6), 1935–1970. Roberts, M.J., Key, N., O’Donoghue, E., 2006. Estimating the extent of moral hazard in
FAO (Food and Agriculture Organization of the United Nations), 2002. Manual on the crop insurance using administrative data. Rev. Agric. Econ. 28 (3), 381–390.
Submission and Evaluation of Pesticide Residues Data for the Estimation of Roll, K.H., 2019. Moral hazard: the effect of insurance on risk and efficiency. Agric. Econ.
Maximum Residue Levels in Food and Feed. Food and Agriculture Organization, 50 (3), 367–375.
Rome. Smith, V.H., Goodwin, B.K., 1996. Multiple peril crop insurance, moral hazard and
Feng, S.Z., Han, Y.J., Qiu, H.G., 2021. Does crop insurance reduce pesticide usage? agricultural chemical use. Am. J. Agric. Econ. 79 (2), 428–438.
Evidence from China. China Econ. Rev. 69, 101679. Walters, C.G., Shumway, C.R., Chouinard, H.H., Wandschneider, P.R., 2012. Crop
Freudenreich, H., Mußhoff, O., 2018. Insurance for technology adoption: an insurance, land allocation, and the environment. J. Agric. Resour. Econ. 37 (2),
experimental evaluation of schemes and subsidies with maize farmers in Mexico. 1–20.
J. Agric. Econ. 69 (1), 96–120. Weber, J.G., Key, N., O’Donoghue, E., 2016. Does federal crop insurance make
Garming, H., Waibel, H., 2009. Pesticides and farmer health in Nicaragua: a willingness- environmental externalities from agriculture worse? J. Assoc. Environ. Resour. Econ.
to-pay approach to evaluation. Eur. J. Health Econ. 10 (2), 125–133. 3 (3), 707–742.
Gomes, H.D.O., Menezes, J.M.C., Costa, J.G.M., 2020. A socio-environmental perspective Wu, B., Pretty, J., 2004. Social connectedness in marginal rural China? The case of
on pesticide use and food production. Ecotoxicol. Environ. Saf. 197 (1), 110627. farmer innovation circles in Zhidan, North Shaanxi. Agric. Hum. Val. 21, 81–92.
Goodwin, B.K., 2001. Problems with market insurance in agriculture. Am. J. Agric. Econ. Wu, Y., Xi, X., Tang, X., 2018. Policy Distortions，Farm Size，and the overuse of
83 (3), 643–649. agricultural chemicals in China. Proc. Natl. Acad. Sci. USA 115 (27), 7010–7015.
Goodwin, B.K., Smith, V.H., 2013. What harm is done by subsidizing crop insurance. Am. Xu, W., Zhang, J.H., 2023. The carbon emission reduction effect of policy-based
J. Agric. Econ. 95 (2), 489–497. agricultural insurance: evidences from the implementation of total cost insurance
Goodwin, B.K., Vandeveer, M.L., Deal, J.L., 2004. An empirical analysis of acreage effects and income insurance pilot program. Insur. Stud. (2), 20–33 (in Chinese).
of participation in the federal crop insurance program. Am. J. Agric. Econ. 86 (4), Xu, L.C., Zhou, L., Pan, S.M., 2013. "One household, two systems" : social self-protection
1058–1077. under food safety threat. Chin. Rural Econ. (5), 32–44 (in Chinese).
He, J., Zheng, X., Rejesus, R.M., Yorobe, J.M., 2020. Input use under cost-of-production Yan, L., Keith, H.C., 2009. A Cost Function Analysis of Crop Insurance Moral Hazard and
crop insurance: theory and evidence. Agric. Econ. 51 (3), 343–357. Agriculture Chemical Use. Milwaukee. Wisconsin Agricultural and Applied
Hill, R.V., Kumar, N., Magnan, N., Makhija, S., 2019. Exante and expost effects of hybrid Economics Association.
index insurance in Bangladesh. J. Dev. Econ. 136 (A), 1–17. Zhang, Z.X., Mu, Y.Y., Hou, L.L., 2018. Does participation in agricultural insurance
Horowitz, J.K., Lichtenberg, E. Insurance, 1993. Moral hazard, and chemical use in optimize factor allocation? An analysis of endogenous farmers’ insurance decision-
agricultural. Am. J. Agric. Econ. 75 (4), 926–935. making and its effect on production. Chin. Rural Econ. 406 (10), 53–70 (in Chinese).
Huang, J.K., Hu, R.F., Pray, C., Qiao, F.B., Rozelle, S., 2003. Biotechnology as an Zhang, Y., Long, H., Li, Y., 2020. How does off-farm work affect chemical fertilizer
alternative to chemical pesticides: a case study of bt cotton in China. Agric. Econ. 29 application? Evidence from China’s mountainous and plain areas. Land Use Pol. 99,
(1), 55–67. 104848.
Jiang, S.Z., Li, L.D., 2021. The analysis of the influence of total cost insurance on Zhong, F.N., Ning, M.X., Xing, L., 2007. Does crop insurance influence agrochemical uses
agricultural economy and social welfare. Insur. Stud. 399 (7), 76–88 (in Chinese). under current Chinese situations? A case study in the Manasi watershed, Xinjiang.
Just, R., Pope, R., 1979. Production function estimation and related risk considerations. Agric. Econ. 36 (1), 103–112.
Agric. Appl. Econ. Assoc. 61 (2), 276–284.
Just, R., Pope, R., 2003. Agricultural risk analysis. Adequacy of models, data, and issues.
Am. J. Agric. Econ. 85 (3), 1249–1256.