台灣房地產波動性分析-不對稱T-GARCH模型的實證

Abstract

系統性的研究房地產市場已經行之有年了，多變數線性迴歸模型，特徵價格函數模型，時間序列模型等。以往多以探討影響房地產價格的因子並且計量化影響力的大小。而在房地產特性的研究方面卻是鮮少有人去討論之。經濟學上的蜘蛛網理論實證了時間序列所擁有的經濟含意，而且時間序列在捕捉波動性的能力也非常精準，所以本研究將使用時間序列模型探討台灣房地產所擁有的特性。以往的文獻已經實證過台北地區房地產市場有波動不對稱性的反槓桿效應。並且許多研究認為房地產市場由於擁有居住性的需求財和投資性的投資財兩種特性，每個城市在這兩種特性有不同的含量，造成配適的時間序列模型不同。所以本研究嘗試使用異質變異時間序列模型對台灣四個都會區進行建構模型。結果發現房地產市場模型有ARMA-GARCH和ARMA-T-GARCH模型等兩種模型。發現了高度成熟的房地產市場台北地區以T-GARCH模型配適度最高。由於台北有T-GARCH性質，本研究最後將使用T-GARCH模型在總體經濟因子上，如果總體經濟因子也擁有T-GARCH模型，也可以嘗試使用多變量T-GARCH模型計量出總體經濟因子對房地產波動性的影響力。總體經濟因子資料選取了11個因子，其中有8個因子也有配適度高的T-GARCH模型。最後本研究組合出了一組多變量T-GARCH模型，而對房地產有負面影響只有房地產自己本身，而消費者物價指數、薪資所得、台灣股價和房屋貸款餘額有正面的影響力。

Systematic study of the real estate market has been decades, such as multivariable linear regression model, hedonic price theory and time series. They must concentrate study in the influence of real estate price by macroeconomic factor and calculate them for the past years. Characteristic of real estate have been talk seldom. We use cobweb theory by economics to prove that time series implied economic meaning, and time series excellent ability to capture the volatility. For the reason, this research tries to use time series to explore characteristic of real estate in Taiwan. They have proved anti-leverage effect in the asymmetric volatility of the Taipei real estate market by previous literature.Many reference thinks that real estate market has two properties, the nature of residential service of demand goods and the nature of investment of investment goods. Every city has different content of the two properties lets them has different suitable time series module.For the reason, this research tried to analysis four metropolitan area of Taiwan by Heteroskedasticity time series model. This research found out there have two suitable time series module, ARMA-GARCH and ARMA-T-GARCH. High maturity of real estate market Taipei suits complex module “T-GARCH”. Then this research will use T-GARCH in macroeconomic factor. If it can be, we will use multivariable T-GARCH model to measure the influence of macroeconomic factor in Taipei real estate price. Eight of the eleven factors are suitable T-GARCH model. Finally, this research finds a multivariable T-GARCH model between house price indx and macroeconomic factor. There have negative influence by house price index itself and positive influence by CPI, salary,TWII and the balance of hosing loan.