中古屋與法拍屋市場交易的房屋性質相近，家戶因自住與投資需求而在二手成屋市場的買賣行爲，將影響此兩市場的供給與需求，進而改變兩市場的房屋均衡價格，所以二者間應存在相關性。目前台灣關於房價的研究少見納入法拍屋市場的議題，本文因此針對兩市場的關係進行探討。台灣房地產市場的價格具多變性且存在結構性轉變，據此，傳統誤差修正模型將無法合理描繪出中古屋價格與法拍屋拍定價格的價格調整行爲與領先落後關係，爲合理與完整分析此兩變數非線性價格調整行爲，本研究以Hansen and Seo (2002)的門檻誤差修正模型進行此議題的探討。由結果發現，中古屋價格指數與法拍屋拍定價格指數間的確具有長期均衡關係，且門檻誤差修正模型的解釋能力相較於傳統誤差修正模型爲佳，因此門檻誤差修正模型的估計結果應能提供更具參考性的資訊。在門檻值前後，變數對於誤差修正的調整速度的確有所不同，變數之間的領先及落後關係亦有所差異。以文中代表性狀態的估計結果推論，這兩市場的價格應互爲領先落後的關係。但不論在哪個門檻階段，皆呈現法拍屋拍定價格指數的調整速度高於房屋價格指數的調整速度，故推論其可能原因爲投資客對資訊快速反應的買賣行爲，使得法拍屋市場進行誤差修正是較有效率的。

Because the traded houses in the housing market are similar to that in the court auctioned housing market, the household's living and investing demand for the houses will simultaneously affect the supply and demand of houses in these two markets, and then that leads to changes in the housing equilibrium prices. Based on this viewpoint, one may infer that the housing prices in these two markets are correlated. However, few researchers investigate the housing market taking the court auctioned housing market into account. This article, thus, intends to analyze the relationship between housing prices and auctioned housing prices. Some housing studies demonstrated that Taiwan's housing market have the changeable property and structural change. Using the tradition error correction model may not reasonably explain Taiwan's housing price behaviors. In order to more appropriately analyze the nonlinear relationship between housing prices and auctioned housing prices, this study analyzes Taipei's housing market by using Hansen and Seo (2002) threshold model. Since Hansen and Seo's model can estimate the unknown threshold level, we first use this model to directly estimate the threshold level, and then to discuss whether the adjustment behaviors and relationships of these two variables are the same in different disequilibrium regimes. According to the results, one can find that the long-term relationship exists between the housing prices and auctioned housing prices. The price behavior can be better explained by the threshold model than by the traditional error correction model. At the different threshold regimes, the adjustment speeds and lead-lag relationship are different. Because the typical regime in our study is the main situation that the error-correction of these two variables occurs, we conclude that these two markets have two-way lead-lag relationship. Moreover, the adjustment speed for the error correction is faster in the auctioned housing market than that in the housing market. The reason, which leads to efficient price-adjustment in the auctioned housing market, may be the investors behavior of fast responding to the market information.