基于GARCH与Markov转换模型度量房价波动风险

为了防范房地产价格快速上涨或下跌带来的不利影响,监测房价波动风险具有重要的意义。该文采用自回归条件异方差(GARCH)模型和Markov机制转换模型研究房价波动过程。首先依据4个中国城市的数据对2类模型分别进行回归,然后根据回归结果计算房价波动率和房价下行概率,以此衡量房价的波动风险。结果显示:Markov模型的样本内拟合程度略高于GARCH模型,但这一差别十分有限。在预测房价下行概率时,Markov模型给出的结果对房价在短期内的变化情况较为敏感,而GARCH模型的预测值较为保守。虽然依据2种模型测算的房价下行概率在数值上有差别,但其走势基本一致,可以为投资者提供参考。

Measuring housing price volatilities:A comparison between GARCH and Markov switching models

In many situations,measuring the uncertainty of housing price fluctuations is an important issue.This study employed two types of univariate time-series models,GARCH and Markov Regime Switching,to study the housing price dynamics in four Chinese cities.Both housing price votatility and the probability of downturns were used to evaluate the housing price uncertainty for each model.The two models have very limited differences for the in-sample fitness.However,the out-of-sample forecasts differ greatly.This paper shows that the Markov Regime Switching model tends to make over-confident predictions while the GARCH model is relatively conservative.Nevertheless,the trends of both risk measurements from the two models were consistent with each other,providing valuable information to investors.