农户正规融资信用风险的度量研究

发布时间：2018-08-12 10:58

【摘要】：从2004年起，中共中央连续多年发布关于农业、农村问题的1号文件，表明党和国家解决“三农”问题的决心。金融机构在农村金融市场的投入偏低，农户开展农业产业化、现代化资金不足，是制约农村建设、农业发展、农民增收的突出问题。要解决这一突出问题，主要依靠农村金融的资金支持。然而，目前农村金融市场面临着许多突出问题：农户信用体系尚未建立起来，农户在信贷业务中由于拥有较多的信息而处于有利地位，导致农户道德风险和逆向选择行为时有发生，金融机构开展农户信贷业务面临较大的信用风险，不良贷款率居高不下，金融机构不愿进入农村金融市场造成市场上供给主体偏低。因此，识别和管理农户信用风险就有十分重要的意义。但是目前我国金融机构对农户信用风险的度量和评价仍处于主观性很强的古典信用定性的分析阶段，主要依靠信贷员的工作经验，金融机构对农户信贷业务缺乏有效的信用风险控制手段。针对上述问题，本文以农户正规融资的信用风险为研究对象，，探索适合度量我国农户信用风险的模型或方法，降低农户违约风险，提高金融机构进入农村市场、开展以农户为服务对象信贷业务的积极性。本文首先通过回顾信用风险度量模型的发展历程，重点介绍了四个信用风险度量模型，为度量农户正规融资信用风险度量提供了可以选择的模型类型。其次对农户正规融资信用风险度量的起因进行了分析，农户有着旺盛且多元化的融资需求，但是金融机构开展农户信贷业务的运营成本较高，对农户信贷风险缺乏有效的控制手段，造成针对农户开展的信贷业务供给偏低，无法满足农户日益增长的资金需求。再次，介绍了农户的信用风险有别于普通贷款的信用风险较高的原因，对农户正规融资信用风险的独特性进行了具体的分析，指出我国农户信用风险度量现阶段最可行的方法是多元统计分析方法。然后，在实地调研数据的基础上对农户正规融资信用风险度量进行实证分析：将影响农户信用风险的指标设计为家庭人口特征、家庭财富拥有量、借贷因素三类25个指标，将这些指标分别输入到判别分析模型和Logistic回归分析模型中，得出如下结论：判别分析模型更倾向于选择逐步判别分析模型，根据自变量对识别农户信用风险贡献的大小，有欠款总额、土地质量、贷款年利率、过去12个月的农业总支出、耕地面积、外出务工人数、农业劳动力人数、资产价值、家庭规模、是否是小组担保、农户联保的成员、65岁以上老人数、是否是信用社成员、存款比例、过去12个月的消费总支出、12岁以下儿童人数、户主受教育程度、信誉评价、过去12个月的外出务工收入18个变量依次进入模型，模型对农户正规融资信用风险的判断的准确率为88.5%。Logistic回归分析模型倾向于选择使用向后逐步法，当Logistic回归分析到第12步时对农户正规融资信用风险综合识别的正确率为84.3%。通过实证分析可以看出，逐步判别分析模型在对农户信用风险识别和评价的准确率上高于Logistic回归分析模型，逐步判别分析模型能够成为金融机构控制农户信用风险的有效手段。最后，为了缓解农村金融市场上的供需矛盾，为了使逐步判别分析模型作为农户正规融资信用风险控制的有效手段得以推广，给出相应的政策建议。
[Abstract]:Since 2004, the Central Committee of the Communist Party of China has issued the No.1 document on agriculture and rural issues for many years, which shows the determination of the Party and the state to solve the "three rural" problem. However, the rural financial market is facing many outstanding problems: the peasant household credit system has not been established, the peasant household is in a favorable position because of having more information in the credit business, leading to the peasant household moral hazard and adverse selection behavior occurring from time to time. Financial institutions are facing great credit risks in developing peasant households'credit business, the rate of non-performing loans is high, and the reluctance of financial institutions to enter the rural financial market results in the low supply subject. Therefore, it is of great significance to identify and manage peasant households' credit risks. The evaluation is still in the stage of classical credit qualitative analysis with strong subjectivity, mainly relying on the work experience of the creditors, and the financial institutions lack effective means to control the credit risk of farmers'credit business. The model or method can reduce the default risk of peasant households, improve the enthusiasm of financial institutions to enter the rural market and develop credit business for peasant households.
Firstly, by reviewing the development of credit risk measurement model, this paper introduces four credit risk measurement models, which provide alternative models for measuring the credit risk of farmers'formal financing. However, the operation cost of farmers'credit business in financial institutions is high, and the credit risk of farmers is lack of effective control means. As a result, the supply of credit business for farmers is too low to meet the growing demand of farmers. Thirdly, the paper introduces the credit risk of farmers is different from that of ordinary loans. This paper analyzes the uniqueness of farmers'credit risk in formal financing and points out that the most feasible method to measure farmers' credit risk is multivariate statistical analysis. The risk indicators are designed as 25 indicators of household demographic characteristics, household wealth ownership, and lending factors. These indicators are input into the discriminant analysis model and the logistic regression analysis model respectively. The conclusion is that the discriminant analysis model is more inclined to choose the stepwise discriminant analysis model and identify the credit risk of farmers according to the independent variable pairs. Contribution size, total amount of arrears, land quality, annual interest rate of loans, total agricultural expenditure over the past 12 months, arable land area, number of migrant workers, number of agricultural labor force, asset value, family size, whether it is group guarantee, members of farmers'joint insurance, number of people over 65 years old, whether it is a member of credit cooperatives, deposit ratio, the past 12 months of consumption The total expenditure, the number of children under 12 years old, the educational level of the household head, and the credit rating of migrant workers in the past 12 months entered the model in turn. The accuracy of the model was 88.5%. Logistic regression analysis model tended to use backward stepwise method when Logistic regression analysis reached 1. Through the empirical analysis, it can be seen that the accuracy of the stepwise discriminant analysis model is higher than that of the logistic regression analysis model in identifying and evaluating farmers'credit risk. The stepwise discriminant analysis model can be an effective tool for financial institutions to control farmers' credit risk. Finally, in order to alleviate the contradiction between supply and demand in rural financial market, the corresponding policy suggestions are put forward in order to popularize the stepwise discriminant analysis model as an effective means to control the credit risk of farmers'formal financing.
【学位授予单位】：西北农林科技大学
【学位级别】：硕士
【学位授予年份】：2012
【分类号】：F832.35;F224

【参考文献】