| 洪源,王群群,苏知立.地方政府债务风险非线性先导预警系统的构建与应用研究[J].数量经济技术经济研究,2018,(6):95-113 |
| 地方政府债务风险非线性先导预警系统的构建与应用研究 |
| Construction and Applied Research of Local Government Debt Risk Nonlinear Early-warning Systems |
| |
| DOI： |
| 中文关键词: 地方政府 债务风险 预警系统 噪声信号比 神经网络 |
| 英文关键词: Local Government Debt Risk Early Warning System Noise Signal Ratio Neural Network |
| 基金项目:本文获得教育部人文社会科学研究青年基金项目“我国农村普惠金融发展的空间差异及调控对策研究”（15YJC790011），以及山东省社科联人文社会科学课题、山东经济社会发展委托课题"振兴山东实体经济推进供给侧结构性改革研究"（17-JS-12）的资助。 |
|
| 摘要点击次数: |
| 全文下载次数: |
| 中文摘要: |
| 研究目标：构建地方政府债务风险非线性先导预警系统。研究方法：运用TOPSIS-AHP法和K-均值聚类法测算我国2006~2016年30个省份地方政府债务综合风险输出评估样本值，运用噪声信号比法进行预警输入指标的精简筛选，然后将预警输入和输出样本数据导入GA-BP神经网络中开展债务风险预警的训练和检验。研究发现：将前置一年的预警输入指标样本值输入构建的GA-BP神经网络中，预警系统的风险输出值平均误差为-0.0375，输出风险状态区间判断的准确率为93.33%，构建的债务风险预警系统具有较强的精确性和先导预警性。研究创新：借鉴“非线性先导法”的预警思想，从债务运行的“举借—使用—偿还”环节出发，基于风险链视角设计了地方政府债务风险预警指标体系，在此基础上，利用软计算方法集成技术来构建符合地方政府债务风险非线性本质的先导预警系统。研究价值：运用构建的债务风险非线性先导预警系统开展我国“十三五”时期地方政府债务风险预测评估，据此提出有针对性的地方政府债务风险的防范化解建议。 |
| 英文摘要: |
| Research Objectives:The construction of local government debt risk non-linear pilot early warning system. Research Methods: The TOPSIS-AHP method and K-MEANS clustering method are used to measure the output sample values of local government debt risk, the noise signal ratio method is used to streamline the parameters of the early warning input. The early warning input and output sample data are imported into the GA-BP neural network of the core early warning model to carry out the training and inspection of debt risk early warning. Research Findings: The first year of early warning input is entered into the constructed GA-BP neural network. The mean error of the risk output value of the early warning system is -0.03756 and the accuracy of the output risk state interval is 93.33%,so the establishment of the debt risk early warning system has the full accuracy. Research Innovations: Drawing the warning idea of “nonlinear lead method”, beginning form the “borrowing-use-repayment” of the local government debt operation, the index system of the local government debt risk nonlinear pilot is designed. Using soft computing method integration technology to construct an early warning system for the non-linear nature of government debt risk. Research Value: Using early warning system in the local government debt risk, the research carries out the risk forecast of local government debt during the 13th five-year plan period and puts forward effective measures to prevent the risk of local government debt. |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
