1.浙江大学 经济学院, 浙江 杭州 310027
2.宁波大学 商学院, 浙江 宁波 315211

[作者简介] 1.沈满洪(https://orcid.org/0000-0002-0315-5185),男,浙江大学经济学院教授、博士生导师,宁波大学商学院教授,宁波大学东海研究院院长,经济学博士,主要从事资源与环境经济学研究; 2.强朦朦(https://orcid.org/0000-0002-7173-0508),男,浙江大学经济学院博士研究生,主要从事资源经济学研究。

Research Progress in Assessing and Managing Agricultural Production Risks
Shen Manhong1,2, Qiang Mengmeng1
1. School of Economics, Zhejiang University, Hangzhou 310027, China
2. School of Business, Ningbo University, Ningbo 315211, China
Abstract

Agricultural production is a complex process involving both natural and economic reproduction, which often presents strong volatility. The frequent natural disasters and decentralized management make China's agricultural production more complex. Identifying, evaluating, and managing agricultural production risks have always been the focus of researchers and policy managers. A comprehensive summary and review of the relevant theories of agricultural production risks will contribute to its future development.
The connotation of agricultural production risks is derived from the general concept of risk, which refers to the possibility that the actual yield of agricultural products is lower than the expected yield due to various risk factors. The connotation of agricultural production risks needs to be understood from two aspects. On the one hand, the risk factors of agricultural production should not only refer to meteorological disasters, but also include a series of possible yield reduction factors such as pollution, pests, and diseases. On the other hand, in specific evaluation cases, most scholars use probability to quantify the possibility of loss. For pollution, pests, and diseases, however, this may be rather difficult, and only semi-quantitative evaluation can be done with the focus on the size of the expected loss.
There are three kinds of quantitative assessment methods of agricultural production risks: parametric distribution fitting, kernel density estimation, and nonparametric information diffusion modeling. The parametric distribution fitting method has good gradualness but it depends on the subjective hypothesis of a prior distribution. The kernel density estimation method is flexible but it does not perform well in small samples and cannot capture extreme events. The nonparametric information diffusion modeling method is suitable for small sample data but the key parameters of this method do not have any unified standard, and the specific probability density function cannot be obtained.
The risk assessment of agricultural production is practiced at the national and provincial levels. The production risks caused by specific factors such as pollution, diseases, pests, and meteorology, as well the comprehensive production risks under the joint action of multiple factors are all involved in the literature. At the same time, the risk assessment of agricultural production has been widely used in risk warning, risk zoning, financial product pricing, and insurance subsidy policy improving.
In the absence of an effective external risk dispersion mechanism, farmers will adopt such measures as income diversification, production diversification, planting low-risk crops, reducing risk investment, and smoothing consumption to carry out ″self-insurance″. The experience of low-income countries shows that these strategies are not all effective and have potentially high costs while stabilizing production. Moreover, these strategies may be an important explanation for the slow growth of farmers' income and even the danger to fall into the poverty trap they are facing.
As a hotspot in agricultural economics, innovative research results are expected to yield in the domain of the evaluation methods of agricultural production risks, evaluation cases, the effectiveness of management strategies, and the design of intervention mechanisms.

Keyword: agricultural production risks; risk management; risk assessment; poverty trap

【主持人语】 农业生产的自然性、地域的分散性以及农产品较小的需求弹性, 决定了农业相比于其他产业面临着更大的自然风险和市场风险。因此, 与风险相关的研究一向是学界关注的重要领域。沈满洪等的《农业生产风险评估及管理研究进展》一文, 从农业生产风险的内涵、评估方法及其应用领域、管理策略等维度为我们进行了农业生产风险评估的全景呈现, 有助于我们加深对该领域的了解。需要强调的是, 我们在关注、评估自然风险, 尤其是突发自然灾害带来的利益损失的同时, 还需要以更全面的视角考察自然风险与市场风险的内在关联。在互联网迅速发展导致信息传递扩散成本大幅下降的背景下, 自然风险诱发的市场风险以及二者的叠加往往使农户面临“ 双重打击” 。因此, 在风险管理工作中, 切勿顾此失彼。周洁红等的《危机背景下信息干预对认证猪肉的信任水平与支付意愿的影响研究》一文, 以非洲猪瘟这一突发危机事件为背景, 分析了信息干预对消费者对认证猪肉信任水平和支付意愿的影响, 其研究结论具有非常重要的启示。在农业风险管理与干预的实践中, 我们可以尝试及时披露信息, 提升农产品品质, 实现品牌化经营, 进而增强消费者信心并提高产品溢价水平, 最终使农户在应对自然风险与市场风险中实现损失最小化。

(一) 单产趋势

(二) 分布建模

1.参数分布拟合法

2.核密度函数估计法

$fh(x)=1nh∑i=1nkx-xih$(1)

$minhMISE(h)=E∫(fh(x)-f(x))2dx$(2)

h=1.06σ $n-15$(3)

h=0.9× min $标准差, 四分位矩1.34$× $n-15$(4)

$f(x)=1nhλi∑i=1nkx-xihλi$(5)

λ i= $f(xi)'G-α$(6)

3.非参数信息扩散法

f(ui)= $1h2π$exp $-(y-ui)22h2$(7)

h= $1.6987(b-a)/(m-1) 1< m≤51.4456(b-a)/(m-1) 6≤m≤71.4230(b-a)/(m-1) 8≤m≤91.4208(b-a)/(m-1) m≥10$(8)

(三) 风险表达

(一) 农业生产风险的评估

1.特定因子造成的生产风险

2.多重因子造成的综合生产风险

(二) 农业生产风险评估的应用

1.风险预警

2.风险区划

3.金融产品定价

4.完善保险补贴政策

(一) 收入多样化

(二) 生产多样化

(三) 种植低风险作物

Decron建立了一个资产积累的经济模型以说明不同收入水平农户的作物选择。基于坦桑尼亚干旱地区的农户数据, 他发现为应对生产风险, 最富有的家庭将不到2%的土地用于种植低风险低回报的甜土豆作物, 但最贫穷的家庭会为此分配9%的土地。这导致最富裕群体的作物组合的平均回报率比最贫困群体高出25%[72]。Kurosaki和Fafchamps基于跨期分配模型, 发现巴基斯坦旁遮普省的农户为降低产量和价格风险会提高饲料作物的种植, 减少香米的种植, 这导致了2%的收入减少和9.4%的福利损失[73]

(四) 减少风险性投入

Rosenzweig和Binswanger建立了均值— 方差效用最大化模型来阐述风险对农户投资构成的影响, 发现印度的农户为应对降雨量波动带来的生产风险会降低生产性投入的比例。降低一个标准偏差的天气波动将使最贫困农户的平均利润提高35%[75]。Rosenzweig和Wolpin发现, 公牛虽然对印度农户的生产有着重要的促进作用, 但为预防风险, 公牛的投资是不足的[76]。Lamb构建了风险规避型农户的两期决策模型, 发现印度半干旱热带地区的农户会因为天气风险而减少化肥的使用, 但在有非农收入来源的情形下, 化肥的需求将会增多[77]。而且, Zimmerman和Carter的理论模型表明, 投资策略是财富的分段函数, 贫困家庭为应对风险会减少风险资产的积累, 这使其陷入长期贫困的恶性循环而无法自拔, 即所谓的贫困陷阱[78]。Dercon和Christiaensen基于埃塞俄比亚的农户面板数据, 发现降雨风险会显著抑制贫困农户对化肥的使用, 使农户持续贫困[79]

(五) 平滑消费

(一) 农业生产风险的评估方法有待完善

(二) 农业生产风险的评估案例有待丰富

(三) 我国农户生产风险的管理策略及有效性分析有待开展

(四) 农业保险对农户风险管理策略的影响有待分析