Research On The Value At Risk OF Real Estate Price Based On GARCH Family Models And Quantile Regression

Over the past ten years, with the rapid development of China’s overall economy, the real estate industry has been an unprecedented boom, causing widespread concern. As the real estate industry and many other industries are highly relevant, driven by strong, to maintain its healthy and stable development of the demand is increasingly prominent, and the real estate market is a vital problem in the real estate market. With a too high price level or a too fast growth rate of the real estate, it will accelerate the formation of real estate bubble, and cause financial risks; While housing prices are too low or too much impact on sales, it will result in a developer funding strand breaks, assets shrinking of the property owner’s, increasing the risk of bad debts of banks and affecting the economy and the people’s livelihood. As a kind of "huge consumer goods" and a product with increasingly mature value preserving and increasing, the real estate price risk has caused more and more attention, and the corresponding risk assessment methods are becoming more and more urgent to be solved.As an effective risk measurement method, the method of VaR is widely used in financial assets risk measurement which has an intuitive method, an easily quantified result, and an easy conclusion to be understood. This paper tries to apply it to the real estate market in the risk of housing prices. While value at risk is a quantile in the risk value itself, the quantile regression method can more accurately simulate the distribution of the financial sequence with a skewness characteristic, so we can get a more comprehensive analysis from it. So this paper chooses to establish a quantile regression model and GARCH model to analyze the risk of house price risk, which has compared the advantages and disadvantages of the two kinds of models, selected an appropriate method to measure the true risk of housing prices, and finally it provides some policies recommendations to investors and government policy makers corresponding to the analysis.The main structure of the article is as follows:Firstly, through an introduction of the first chapter, the paper introduces the background and the research status of VaR theory and quantile regression at home and abroad. Then in the second chapter, it mainly introduces the principle of VaR, the theory of GARCH models, and the theory of quantile regression technique, which provide theoretical support for the following empirical research. The third chapter is the core part of this thesis, an empirical study on real estate price in Zhengzhou City. This chapter based on Zhengzhou City real estate market uses Zhengzhou new housing price index to calculate the value at risk, and the GARCH models of EGARCH and GARCH are applied based on the normal distribution, t distribution and generalized error distribution. In order to compare differences between the two models in the estimation of the value at risk of housing and choice an appropriate model from them, the the fourth chapter compared two aspects of the volatility (mean relative deviation index and the average percentage error) and the accuracy (Kupiec maximum likelihood method) of results, after calculating the value at rist of the new housing price index. The fifth chapter is about the conclusion and policy recommendations.In this paper, the following conclusions are drawn:the estimated value of the risk value of quantile autoregression model is the most close to the average level of all models’ estimations, which reflected its advantages in building model with no assumption of series distribution for quantile regression. The precision of measurement for the GARCH models depends on the choice of the appropriate distribution assumption, and the normal distribution is not suitable for describing the distribution of housing prices, and t disturibution assumption will get an over estimated result of the actual risk, while GED distribution has a better estimation of the actual risk. The estimation of value at risk based on GARCH models will be different at a high confidence level. At the same time, the study of prices in Zhengzhou has found that the return rate serie of Zhengzhou City new residential price index is skewed to the left characteristics, and there exists the volatility clustering effect, affected by the "leverage", which represents as easily to rise and difficultly to fall. Return value to the previous is greatly influenced by the last term, and the previous rate is mainly concentrated in the vicinity of zero when last term at a low level; while the last term is at a higher level, the previous will be at higher too and has a more dispersed distribution. It has a long-term downward trend for return serie with a lesser extent, but the housing prices will still have a rise trend overall.