預售屋大量估價模型之建立

A Mass Appraisal Modeling of Pre-sale Housing

相較於中古屋已發展大量估價系統網站供民眾查詢住家房價，預售屋因具有個案特徵，難以直接適用中古屋模型進行估價。本文納入基差與商譽變數以建立預售屋大量估價模型，實證結果顯示預售屋平均單價與附近成屋價差比為35.5%，基差為23.73萬元，上市上櫃建商相較於非上市櫃建商預售屋價格高出5.7%；模型估計樣本外的平均絕對百分比誤差為11.82%，±10%及±20%命中率分別為48.81%與86.07%。建立預售屋大量估價模型除可供民眾查詢預售屋價格，亦可應用於稽核預售屋價格申報不實的情況，且可藉由長期預測觀察預售屋與成屋市場價格變化，有助於價格資訊流通與穩定市場價格。

Pre-sale housing possesses the property of futures, meaning that prices are affected by the prices of nearby existing housing. Besides, in the case of pre-sale housing traded before the completion of the building, consumers depend on the developer's reputation to evaluate the house's quality. To sum up, the basis and the developer's reputation are important in the appraisal of the pre-sale housing. This paper uses the hedonic method to build an empirical model. The empirical results show that the average price of pre-sale housing in Taipei is 35.5% higher than that of the surrounding existing houses, and for developers that are listed on the stock exchange or in the over-the-counter market, the price is 5.7% higher than that for non-listed developers' housing. In the results for the out-sample model, the MAPE is 11.82%, with a ±10% hit-rate and 48.81% and 86.07% with a ±20% hit-rate. Establishing a mass appraisal of pre-sale housing can provide a search channel for consumers, and also monitor the change between the pre-sale housing market and the existing housing market. It is helpful for information transparency and for stabilizing market prices.