台北地區住宅價格之時間序列特性與模型：結構性時間序列模型之應用

Time-Series Properties and Modelling of House Prices in Taipei Area: An Application of the Structural Time-series Model

本研究採用結構性時問序列模型以分析台北地區住宅價格中不可觀察的時間序列成分。本研究以隨機性與確定性的趨勢成份來展現台北地區住宅價格長期的變動行為。本研究也發現台北地區住宅價格隱含二年與七年的隨機性循環變動成分，雖然台灣住宅市場有七年一循環之說，但此循環不是固定長度的。我們使用這些不可觀察的時間序列成分來估計住宅價格長短期模型，這些模型通過統計檢定並且有不錯的解釋能力，而樣本內與樣本外的預測也都有相當的穩定度。

This paper analyses the unobserved components of Taipei house prices based on the Structural Time-series Model. It shows how stochastic and deterministic trends characterise long-run behaviour of Taipei house prices. Stochastic cycles were found in this price series around 2 and 7 years. Although Taipei housing market is though to have 7-year cycle, model suggests that the cycle is stochastic rather than deterministic. Using statistically specified unobserved components, we tested the long-run and short-run structural time-series house price models. These models all have good forecasting powers and acceptable diagnostic test statistics.