影響中國房地產價格之總體經濟變數分析

The Analysis of Housing Price of China with Macroeconomics Variables

謝煜鴻(Yu-Hung Hsieh)

指導教授 ： 朱家祥

國立臺灣大學/社會科學院/經濟學研究所/碩士(2006年)

https://doi.org/10.6342/NTU.2006.01951

摘要

中國房地產價格之走勢總是吸引中國國內外產、官、學各界人士、甚或是平民百姓的關注，但影響整體房價走勢之因素，尚缺乏實證分析理性而有效的進行討論。本論文採用向量自我迴歸模型（VAR）、ARMA模型及ARMAX模型等計量經濟學分析工具，試圖建立有效解釋房價走勢的模型。本論文分析了房價與利率、貨幣供給、景氣循環、股市與物價間之關係，並分別利用各變數之原始水準項與年增率項進行實證研究，結果採用年增率所建構之模型解釋力遠高於原始水準項；而由年增率模型中本論文獲得以下發現：

1. 根據VAR及Granger Causality Test之結果，本論文發現總體經濟變數中會對房價年增率造成影響的變數有前一期的貨幣供給量年增率、工業增加值年增率及股價年報酬率，影響方向則分別為負向、正向及正向。這代表貨幣供給的增加對房價可能造成負面影響，而在景氣上升階段及股市上漲時，房價則將隨之上漲。另外，房價則Granger Cause工業增加值、物價指數及股價，影響方向則分別為正向、負向及負向；這代表房價的上升也將刺激景氣向上，但會反向壓抑物價及股價走勢。

2. 較為適宜用來描述房價年增率走勢之ARMA模型為ARMA(2,1)，此時模型之配適度R2為82.91%，而將其他變數加入ARMA(2,1)後，配適度可以超越ARMA模型水準者有貨幣供給年增率、工業生產指數年增率、股價報酬率及物價指數年增率，但提高之配適度則相當微小。在本論文所檢驗過之所有模型中，若以AIC及BIC準則選取配適度最佳者，則以加入前一期貨幣供給年增率之ARMAX(2,1)模型為最佳，此時配適度R2為84.46%。

3. 本論文所採用之五種總體經濟變數對房價走勢之影響均無法與歷史房價走勢對自身之解釋力相比。由於中國實施住房改革以來，房地產市場正經歷著中國人民購屋置產的強大需求，或許有其他需求面的因素能更有力的解釋驅動房價變化的因素，而尚難以量化指標表達，因此造成了上述結果。

關鍵字

房價 ； 向量自我迴歸模型 ； 單變數時間序列分析 ； 貨幣供給 ； 中國

並列摘要

This research tries to explain the housing price of China with macroeconomics variables. The macroeconomics variables chosen in the research are money supply (M1), the 7-days CHIBOR, the SSE composite Index, value added of industry (VAI) and CPI. Using Granger causality test and time series model including VAR, ARIMA and ARMAX, the thesis comes out with the following results:

1. According to the Granger causality test and VAR model, we found that the 1-month lag money supply, value added of industry and stock price granger cause housing price in terms of YoY. Housing price granger cause value added of industry, CPI and stock price.
2. ARMA(2,1) is the best model to estimate the housing price on China in this research, and the R2 is 82.91% when using this model. We could not find significant improvement of R2 after added any other macroeconomics variables is ARMAX model. The largest R2 appears in ARMAX(2,1) with exogenous variable M1 when explaining housing price, and the R2 is 84.46%.
3. The results of empirical model represent that the macroeconomics variables can’t explain housing price in China better than housing price itself. The reason may be the strong demand of house after the housing policy revolution. There may exist other demand side factor like population statistics, migration or saving rate dominate the housing price despite the macroeconomics variables.

並列關鍵字

VAR ； ARIMA ； ARMAX ； Granger Causality ； Housing price

參考文獻

Darrat, A. F. and J. L. Glasock (1993),On the real estate market efficiency, Journal of Real Finance Economics, Vol. 7, 1993, 55-72

Goldstein and Nelling (1999) , REIT Return Behavior in Advancing and Declining Markets, Real Estate Finance, 1999, 15, 68–77.

Gordon de Brouwer (2004), Asset Price in Japan, Comment on ‘It Takes More than a Bubble to Become Japan’ by Adam Posen

Hiro Y. Toda; Peter C. Philips (1993), Vector Autoregressions and Causality, Econometrica, Vol. 61, No.6, 1367-1393

Harvey, Andrew(1989), The Econometric Analysis of Time Series, The MIT Press