房屋價格決定因素之探討：空間與多層次分析之應用

Abstract

房屋價格近十年來呈現長期上升之趨勢，且不同都會區之間房價差距也逐漸擴大。針對房價之區域分布議題，過去研究受限於房屋實際價格之取得困難，而無法進行完整之估計，我國內政部因而實施「不動產交易實價登錄」，揭露真實房價並減少房屋交易中的資訊不對等。 本研究以內政部「不動產成交案件實際資訊資料供應系統」所提供之房屋交易資料進行分析，資料期間為2012年1月至2014年4月，共納入388,301筆交易於台灣本島333個鄉鎮市區內之交易資料。研究方法分為兩部分：空間分析與多層次分析。空間分析以Moran’s I指標研究房價的空間自相關現象，並以Anselin’s LISA指標分析房價空間聚集現象之熱區與冷區。再以多層次分析，納入房屋個體因素「有停車位」、「有管理單位」、「土地移轉面積」、「是否為住宅區」、「建物移轉面積」、「屋齡」與「隔間多寡」，以及鄉鎮市區的「高等教育人口比例」、「人口密度」及「地區平均所得」，研究區域因素與房價個體因素，對房價解釋力之脈絡效果。 結果顯示房價具有空間聚集之現象，且其熱區與冷區之分布，與區域因素之熱區與冷區分布有類似。多層次分析更指出，區域因素與個體因素，對房價之解釋力具有脈絡效果。

Housing prices have drastically risen in the last ten years, and the price difference among various urban areas rises in Taiwan. The past researches of the distribution of the housing prices are subjected to data availability and result in the incapability of throughout appraisal. Ministry of the Interior, R.O.C has hence recently carried the “Actual Selling Price Property Value Reporting System” in order to reveal correct housing prices and reduce the information asymmetry. This study analyzes the dataset of houses trades acquired form Actual Selling Price Property Value Reporting System. The dataset includes 388,301 transactions which happened in 333 administrative districts from January 2012 to April 2014. The study consists of two parts: spatial analysis and multilevel analysis. In the spatial analysis we apply Moran’s I to test the spatial clustering of housing prices, pointing out the hot spots and cold spots of housing prices by Anselin’s LISA. Besides, in order to examine contextual effect on housing prices, multilevel analysis is adopted. In the multilevel analysis we examine individual factors, such as “parking space,” “management units,” ”land size,” “residential area,” ”age of house,” and ”partitions,” and regional factors, such as “high education,” ”population density, ”and ”local average income.” Our empirical findings show that there exist spatial clustering of housing prices and similar distribution of hot/cold spots between housing prices and regional factors in Taiwan. Also, in the multilevel analysis it shows that both regional and individual factors exhibit contextual effect on housing prices.