中国房产市场预警系统研究

【作者】 李奇昤；

【导师】 李景国；

【作者基本信息】 中国社会科学院研究生院 ， 区域经济， 2012， 博士

【摘要】 2008年，对绝大部分人来说毫无预警的房地产次贷危机引发了美国的经济危机，不仅使世界第一大经济体---美国的经济大厦飘摇动荡，也对包括中国在内的全球经济造成了巨大冲击，房地产风险对经济造成的惊心动魄的危害由此可见一斑。随着房地产业对中国经济快速增长影响和作用的增加，中国房地产市场尤其是房产市场的稳定发展越来越受到社会各界广泛关注，更为中央政府所重视。中国政府出台了一系列稳定房价的政策，但大部分的稳定政策同步或滞后于现实房产市场出现的变化和问题，使政策的效能大打折扣。反复出现这种现象的最大原因之一就是没有能够有效预测未来房产市场风险的预警系统。已有一些中国学者采用多种方法试图建立中国房产市场或房地产市场的预警系统，但至今不仅鲜有被广泛接受的房产市场预警系统，而且已有研究中大部分都侧重于建立区域性的房产市场预警系统。因此，本文采用到目前为止在中国尚没有研究者采用的方法，即信号分析法和Probit模型，建立了中国全国性的房产市场预警系统。本文基于信号分析法与Probit模型，使用从1999年第一季度至2010年第三季度的宏观经济变量和房地产市场变量构建了中国房产市场预警系统，并对其预测能力进行了检验。本文共由六章构成：第一章，为绪论；第二章，介绍了有关房产价格决定模型、房产价格与宏观经济的关联性；第三章，通过相关关系分析法和Granger因果关系检验，分析了房产价格与宏观经济因素、微观经济因素之间的关系；第四章，采用Granger因果关系、VECM模型的脉冲响应函数分析和方差分解分析，研究了中国三种房产价格指数之间的关系和互相影响趋势；第五章，通过VAR模型分析了影响房产价格的宏观经济因素和房地产市场因素；第六章，以前几章的分析结果为参考，基于信号分析法与Probit模型建立了中国房产市场预警系统，并对其预测能力进行了检验。与之前的有关房产市场预警系统的大部分的研究只进行样本内预测不同，本文的第六章在基于信号分析法和Probit模型建立了符合中国国情的房产市场预警系统后，不仅检验预警系统对过去风险期间的预测和反应能力，即进行样本内预测检验（In Sample Forecast Test），而且进行了样本外预测检验（Out of Sample Forecast Test）。样本内预测检验结果显示，信号分析法的房产市场预警系统在风险发生一年之前的预测能力达到93%，而Probit模型的房产市场预警系统在风险发生一年之前的预测能达到88.3%。样本外预测检验结果显示，信号分析法的房产市场预警系统在风险发生一年之前的预测能力达到90%，而Probit模型的房产市场预警系统在风险发生一年之前的预测能达到80%。因为中国房产市场的历史较短，可用数据有限，因此虽然本文所构建的房产市场预警系统检验预测能力很高，对尚难确定对未来可能发生的市场的风险是否会也具有如此高的预测能力，还需要随时间推移继续观察房产市场的变化而适当调整预警系统的预警指标、临界值、权重等内容。但是，因为本文构建房产市场预警系统和检验其预测能力时，在方法上和过程中没有逻辑错误和伪造，笔者认为可以期待本文所构建的基于信号分析法与Probit模型的房产市场预警系统作为中国全国范围的房产市场预警系统可以发挥良好的作用，为政府采取先行的房产市场稳定化政策、防范市场风险提供科学依据。更多还原

【Abstract】 During the last global financial crisis that originated in the United States, people onceagain realized that the shocks caused by the housing market crisis not only seriously damagedthe entire economy of the United States, the world’s biggest economy, but also had impactsall around the world. That is, in today’s globalized economy, the serious instability andshocks of a nation’s housing market can affect the entire world. Considering that the worldhas not fully recovered from the shocks of the global financial crisis and European fiscalcrisis yet, there is a huge possibility that global economy will face a serious situation that willmake recovery even harder if the housing market in China, which has risen to the secondbiggest economy in the world, falls in danger. That is why so many countries and researchinstitutions are paying attention to the movements of China’s housing market. The timedesperately calls for a need to set up an early warning system for the housing marketappropriate for China’s situation.In fact, many Chinese scholars have long tried to establish a housing market earlywarning system for China in various ways, but there is no housing market early warningsystem that is universally accepted yet. Many of the previous studies focused on setting up ahousing market early warning system centered around certain regions. Thus this study set outto establish a nationwide housing market early warning system for China with twoapproaches that had not yet been applied to the country, namely the Signal Approach Modeland Probit Model. Based on those two models, the investigator set up a housing market earlywarning system for China by making use of the quarterly macroeconomic indexes and realestate market indexes from the first quarter of1999to the third quarter of2010.The paper consists of a total of six parts: Chapter1is the introduction. Chapter2offers atheoretical consideration of house pricing models and connections between house prices andmacroeconomy. Chapter3examined relations between house prices and macro-and micro-economic factors with the Correlation Test and Granger Causality Test. Chapter4analyzedrelations and trends of mutual influence among China’s three major house pricing indexeswith Granger Causality Test and VECM Model’s impulse response analysis and variancedecomposition. Chapter5investigatedthe factors affecting house prices in macroeconomic factors and real estate market factorsseparately with VAR analysis and compared the results. Finally, Chapter6set up a housingmarket early warning system for China with the Signal Approach Model and Probit Model based on the results from the previous chapters, as well as tested its predictability.Unlike most of the previous studies that were carried out in sample forecast tests to testpredictability rather than out of sample forecast tests, the study divided Chapter6into twosections: Section1set up a housing market early warning system proper for the uniquesituations of China with the Signal Approach Model and Probit Model, compared it with thepast risk sections of China’s housing market, and tested the system’s predictability andresponsiveness to the old risks. In a word, Section1conducted an in sample forecast test.Section2conducted an out of sample forecast test with the established housing market earlywarning system.The results of the in sample forecast test show that the Signal Approach Model-based housingmarket early warning system recorded93%accuracy in predicting a risk one year prior torisk occurrence and that the Probit Model-based housing market early warning systemrecorded88.3%accuracy in predicting a risk one year prior to risk occurrence. The results ofthe out of sample forecast test reveal that the Signal Approach Model-based housing marketearly warning system recorded90%accuracy in predicting a risk one year prior to riskoccurrence and that the Probit Model-based housing market early warning system recorded80%accuracy in predicting a risk one year prior to risk occurrence.It is inevitable that the data and information used to set up a housing market earlywarning system had limitations, given that China’s housing market has a short history andlimits to usable data. Thus it is difficult to say that the housing market early warning systemestablished in the study will record as high accuracy as the study in predicting other futurerisks. It is required to make ongoing observations of changes to the housing market,determine a leading composite index of housing market early warning systems according tochanges, and make proper adjustments to the threshold and weight.Since there was no logical deficiency or artificial manipulation in the methods andprocesses of setting up a housing market early warning system and testing its predictability inthe study, the housing market early warning system will be able to serve valid and meaningfulpurposes as a nationwide housing market early warning system in China and provide theChinese government with rational and scientific grounds for deciding a housing market policy.更多还原