基于房地产与金融风险复杂交互行为的风险预警研究

Real estate related financial risk problem has always been an important issue in the financial field, which has gained widespread attention from domestic and foreign experts. However, more research is based on statistics or econometrics, and only consider the real estate market or financial market factors, ignoring the effect of local government and network public opinion. With the rapid development of information science, the real estate related data and financial data are more and more rich. Therefore, this project will start from these massive data to explore intelligent warning in the background of big data. Firstly, the fuzzy clustering method based on particle swarm optimization is utilized to determine the real estate related financial risk key factors. Then, the propagation dynamic model and the association rules mining are presented to explore the conduction mechanism of real estate related financial risk and the complex interaction between real estate and financial risk. Finally, based on knowledge fusion theory, the Logistics and the SVM are combined to establish real estate related financial risk early warning model, then based on cognitive psychology, the visualization of real estate related financial risk warning is proposed. The research results will offer a theoretical reference to recognize the real estate related financial risks more scientifically and accurately, and provide theoretical and technical support for regulators to understand the operation condition of real estate market more efficiently and accurately.

房地产相关金融风险预警问题一直是金融领域的重要课题，受到国内外专家的广泛关注。然而目前较多的研究都是从统计学或是计量经济学角度出发，并且只考虑了房地产与金融市场因素，忽视了地方政府与网络舆论的效应。随着信息科学的迅猛发展，房地产相关数据与金融数据积累的越来越丰富，本项目将从这些海量、多元、异构的数据出发，探讨大数据背景下的智能预警问题。首先运用基于粒子群的模糊聚类方法识别房地产相关金融风险的关键因素；然后运用传播动力模型和关联规则挖掘方法揭示房地产相关金融风险的传导以及房地产与金融风险的复杂交互关系；最后以知识融合理论为指导，运用Logistics和SVM相融合的方法构建房地产相关的金融风险预警模型，并以认知心理学为基础，提出房地产相关金融风险预警的可视化。预期研究成果将为科学准确地认识房地产相关金融风险提供理论参考，并且为管理层高效准确地了解房地产市场运行状况提供理论与技术支持。

自改革开放特别是进入新世纪以来，我国房价经历了持续快速的上涨，造成了资源配置扭曲，也积累了一定的风险。房地产相关风险问题一直是金融领域最具挑战性的课题之一，受到国内外专家的广泛关注。本项目致力于从系统科学和复杂科学的角度出发，运用大数据的技术与方法针对房地产的风险度量、交互传导与预警问题进行研究，在一定程度上为监管部门制定相关风险防范措施提供一定的理论支撑与参考价值。本研究主要从如下几个方面展开：（1）运用文献挖掘与文本挖掘的方法从房地产市场内部与外部环境出发，从市场层面、房地产企业、政策因素、金融机构四个方面探讨房地产风险影响因素，并且运用数据挖掘的方法对所构建的房地产风险指标体系进行降维。（2）运用自回归分布滞后模型，针对房地产限购与房价之间的关系进行了研究，从短期与长期两个角度，研究了租售同权和限购政策常用的三大手段对于房价调控的影响。通过大数据技术获取网络搜索数据构建了政策关注度指数，并以建筑面积、年龄、环数等为特征构建了特征价格指数，定量评估了各类政策指数对房价的影响。采用复杂网络与溢出指数方法从静态与动态测度我国房地产风险传导的强度和方向并绘制中国房地产风险热力图，然后探讨了新冠疫情等重大突发公共事件对中国房地产风险的影响。（3）通过构建人工智能模型进行房地产风险度量与预警并进行实证检验与比较。结果表明：合成的房地产风险指数很好的描绘了房地产风险的周期性波动，并且所构建的预警模型较其他模型有更好的性能，预警准确率更高。