seriespropertiesandmodellingofhousepricesintaiwan.docVIP

Time-Series Properties and Modelling of House Prices in Taiwan: An Application of the Structural Time-series Model 台灣住宅價格之時間序列特性與預測模型: 結構性時間序列模型之應用 MING-CHI CHEN (陳明吉) Department of Finance, National Sun Yat-sen University, Taiwan* 國立中山大學財務管理系 KANAK PATEL Department of Land Economy, University of Cambridge, UK 英國劍橋大學土地經濟系 民國八十九年九月 ------------------------------------------------------ * 聯絡人： 陳明吉，助理教授，國立中山大學財務管理系，高雄市(804)蓮海路70號 Tel: 886-7-5252000ext4826, Fax: 886-7-5254899, e-mail: mcchen@finance.nsysu.edu.tw  Abstract The primary purpose of this paper is to analyse the unobserved components in Taiwan house price series based on the structural time-series model. This paper shows how stochastic and deterministic trend characterise Taiwan house price long-run behaviour after the removal of certain substantial structural changes. We also find evidence of some stochastic cycles around 2 and 7 years in the price series. Although there is a myth of 7-year cycle in Taiwan housing market, the model suggests that the cycle is stochastic rather than deterministic. Based on the unobserved components statistically specified, we test long-run and short-run structural time-series house price models, and find that the ex-post forecasts perform reasonably well over the sample period. Keywords: House Prices, Structural Time-series Model, Unobserved Components 中文摘要 本研究採用結構性時間序列模型來分析台灣住宅價格中無法觀察的時間序列成分。在移除樣本期間內結構性轉變的成分後，本研究以隨機性與確定性的趨勢成份來展現台灣住宅價格長期的變動行為。本研究也發現台灣住宅價格隱含二年與七年的隨機性循環變動成分，1. Introduction The primary purpose of this paper is to analyse the nature of trends, cycles and irregular movements in Taiwan house price series based on structural time-series model (Harvey and Todd, 1983)). This model, explicitly based on the stochastic properties, provide most useful framework to present stylised facts of time series and forecasting ability. Traditionally, the estimation of these unobserved components, which is known as signal extraction, has been carried out