分类号密级
U D C 编号
硕士学位论文
学位论文题目：鸡蛋价格波动预警研究
——基于不用预警模型比较
姓名高新亚
学号2020211226
学院经济学院
学位类别：学术硕士□专业硕士□同等学力
学科专业产业经济学
指导教师于爱芝
第二导师
提交论文日期：2023年5月11日
摘要
蛋鸡产业是我国现代畜牧业中重要组成部分，鸡蛋也在我国居民膳食结构中占据
重要地位，是性价比最高的蛋白来源之一。近年来，随着蛋鸡养殖集中度不断提升、
疫病发生率上升、宏观经济环境恶化、政策不确定性提升以及消费需求的多元转变，
蛋鸡产业发展也面临着转型升级。鸡蛋价格也由小幅波动上涨趋势转变为区间大幅震
荡走势，价格不稳定性增加。鸡蛋作为我国居民膳食结构的重要组成部分，其价格剧
烈波动对消费者和生产者均有较大的福利损失。因此，本文以鸡蛋价格作为研究对象，
具体的研究内容与结论如下：
第一，构建鸡蛋价格预警指标体系。价格预警包括警情指标选择、警限确定和警
兆指标体系构建。在警情指标选择上，根据以往文献资料和学者的研究成果，并考虑
到警情指标应为相对值指标，选择构建蛋肉价格比、蛋料价格比、鸡蛋价格波动率、
蛋粮价格比作为待选警情指标，同时以历史上对鸡蛋价格影响较大的突发事件发生时
间点观察待选警情指标波动和鸡蛋价格波动是否一致，选择蛋料价格比作为最终的警
情指标。预警体系中警限的确定是以统计方法的均数原则作为标准，将警情指标的平
均值和标准差作为划分依据，正负一个标准差内期间属于无警区间，正负一个标准差
外、正负两个标准差内区间属于轻警区间，两个标准差外区间属于重警区间。在预警
警兆指标体系构建上，文章从供给方面、需求方面、疫情疾病、经济环境和重大事件
五个警源选择警兆指标，根据指标对鸡蛋价格的作用机制和数据的获取难易程度，本
章选择了17个待选警兆指标，作为构建警兆指标体系的基础。通过时差相关分析法，
本文计算出鸡蛋价格与各待选警兆指标之间的时差相关系数，根据全面性、代表性、
可操作性和灵敏性四大原则，进而确定警兆指标及其领先或滞后期数，构建警兆指标
体系如下：先行指标有蔬菜价格指数和突发事件，一致指标有玉米价格、蛋鸡配合饲
料价格、白条鸡价格、商品代蛋用雏鸡鸡苗价格、羊肉价格和牛肉价格，滞后指标有
农村居民可支配收入和宏观经济景气先行指数。
第二，构建模型进行价格预警。本文选择ARIMA 模型、BP-神经网络模型和支持
向量机模型作为预警模型分别对鸡蛋价格进行预警，根据模型最终预测结果发现，
ARIMA 模型在近期月份（四个月内）预测准确度较高，但是在较远月份预测准确度较
差，适宜作为短期价格预警模型。BP-神经网络和支持向量机均属于机器学习模型，在
预测准确度上，支持向量机模型高于BP-神经网络模型，并且支持向量机模型错误预
警与其实际预警差异较小，模型稳定性较好，在短期和长期均更适合作为鸡蛋价格预
警模型。
第三，给出政策建议。基于鸡蛋价格预测和波动预警分析，本文认为政府应该建
立蛋鸡产业公共信息服务平台，及时发布蛋鸡产业相关数据和信息。在此基础上，蛋
鸡养殖企业应该充分利用相关数据，针对不同影响下鸡蛋价格的变动情况，逐步完善
鸡蛋价格波动预警方案，并且使用多种手段逐步增强自身的抗风险能力。
本文主要有以下创新点：（1）首次使用BP-神经网络模型进行鸡蛋价格预警。BP-
神经网络在农产品价格预警上有着广泛应用，有良好的预警准确度，但是人工神经网
络需要大量的数据样本才能达到较好的效果，因此还没有神经网络在鸡蛋价格预警方
面的应用研究，本文搜集了2000年1月-2022年12月相关数据，满足使用BP-神经网
络模型样本量要求。（2）对鸡蛋价格的长期趋势和短期波动分别构建预警模型。本研
究使用ARIMA 模型对鸡蛋价格短期波动进行预警，使用BP-神经网络模型和支持向量
机模型对鸡蛋价格进行整体预测，对模型的预警情况进行比较，选择出不同时间跨度
较为合适的模型。
关键词：鸡蛋价格，ARIMA 模型，BP-神经网络模型，支持向量机模型
Abstract
The egg industry is an important part of China's modern livestock industry,and eggs also
occupy an important position in the dietary structure of Chinese residents,being one of the
most cost-effective sources of protein.In recent years,with the increasing concentration of
egg farming,rising incidence of epidemics,deteriorating macroeconomic environment,rising
policy uncertainty and diversified changes in consumer demand,the development of the egg
industry is also facing transformation and upgrading.Egg prices have also changed from a
small fluctuating upward trend to a large oscillating trend in the range,with increased price
instability.As an important part of the dietary structure of the Chinese population,dramatic
price fluctuations in eggs are associated with large welfare losses for both consumers and
producers.Therefore,this paper takes egg prices as the object of study,and the specific
research contents and conclusions are as follows:
First,a system of egg price warning indicators is constructed.The price warning includes
the selection of warning indicators,the determination of warning limits and the construction
of the warning indicator system.In the selection of warning indicators,based on previous
literature and scholars'research results,and considering that warning indicators should be
relative indicators,we chose to construct egg-meat price ratio,egg-feed price ratio,egg price
volatility,and egg-grain price ratio as the selected warning indicators,and observed whether
the fluctuations of the selected warning indicators were consistent with the fluctuations of egg
prices at the time of historical emergencies that had a large impact on egg prices.The egg-to-
food price ratio is used as the final warning indicator.The determination of the alert limits in
the early warning system is based on the mean principle of statistical methods,and the mean
and standard deviation of the alert indicators are used as the basis for division.The period
within plus or minus one standard deviation belongs to the no-alert range,the period outside
plus or minus one standard deviation,and the period within plus or minus two standard
deviations belongs to the light-alert range,and the period outside two standard deviations
belongs to the heavy-alert range.In the construction of the early warning indicator system,the
article selects warning indicators from five sources:supply side,demand side,epidemic
disease,economic environment and major events.17warning indicators are selected in this
chapter as the basis for the construction of the warning indicator system according to their
mechanism of action on egg prices and the ease of data acquisition.Based on the four
principles of comprehensiveness,representativeness,operability and sensitivity,the warning
indicators and their leading or lagging periods are determined,and the warning indicator
system is constructed as follows:the leading indicators are vegetable price index and
unexpected events,the consistent indicators are corn price,egg feed price,white chicken price
The first indicator is the price of maize,the price of egg feed,the price of white chickens,the
price of commercial egg chicks,the price of mutton and the price of beef,and the lagging
indicators are the disposable income of rural residents and the macroeconomic prosperity
index.
Second,the model is constructed for price warning.The ARIMA model,the BP-neural
network model and the support vector machine model were selected as early warning models
for egg prices,and the final prediction results showed that the ARIMA model was more
accurate in the near future (within four months),but less accurate in the far future,making it
suitable as a short-term price warning model.In terms of prediction accuracy,the support
vector machine model is more accurate than the BP-neural network model,and the error
warning of the support vector machine model is less different from the actual warning,which
makes the model more stable and more suitable as an egg price warning model in both the
short and long term.
Third,policy recommendations are given.Based on the analysis of egg price prediction and
early warning,this paper suggests that the government should establish a public information
service platform for the egg industry and release data and information about the egg industry
in a timely manner.On this basis,egg breeding enterprises should make full use of the
relevant data to gradually improve their egg price fluctuation early warning programs in
response to the changes in egg prices under different influences,and use various means to
gradually enhance their own risk resistance.
The main innovations of this paper are as follows:(1)the first use of BP-neural network
model for egg price warning;BP-neural network is widely used in agricultural price warning
and has good warning accuracy,but artificial neural network needs a large number of data
samples to achieve better results,so there is no research on the application of neural network
in egg price warning,and this paper collects the data from 2000This paper collects relevant
data from January 2000to December 2022,which meets the sample size requirement of using
BP-neural network model.(2)An early warning model was constructed for the long-term