| 摘要: | 摘 要
於西元1992年8月原能會偵測發現，全台灣地區有1,659棟輻射屋，除了會影響輻射屋居民的生命及財富價值，還可能對鄰近地區產生負面影響，故本研究擬探討輻射屋對鄰近房屋價格是否產生影響，與距離不同所造成房價影響之差異程度。而資料來源根據台北市與新北市於1993年至1996年台灣不動產成交行情公報之「交易資料」，利用地理資訊系統(GIS)軟體，計算與輻射屋之距離，並且以特徵價格法(hedonic price method)來探討。
其架構分為：半對數線性及非線性(Box-Cox)特徵價格方法，並且個別再分為兩種模型：第一、分級距離間距；第二、直線距離間距等模型，台北市刪除1000公尺以上，而新北市刪去相距高於800公尺的樣本，來進行實證，其結論可歸納如下：
利用半對數線性模型，除了台北市模型(一)分級距離間距之配適度(39.67%)
較低之外，其餘都約有60%以上，顯示半對數線性模型對於房屋價格已有相當高的解釋能力。而利用Box-Cox函數轉換，無論在台北市或新北市的實證結果中，所得到的θ之轉換係數，在1%統計水準下，具有統計顯著異於零，因此兩者模型都有相當高的解釋能力。
二、房屋特徵變數之實證結果符合本研究預期，在係數符號方面，住宅面積、房屋類別三(透天、別墅)等，皆對於房價具有顯著正相關，而銷售天數與類別二(指公寓)對於房價為負面影響。另外，其餘房屋屬性自變數對於房價皆具有顯著影響。
三、在距離方面，台北市與新北市兩者的研究結果中，基於與輻射屋相距低於10公尺的標準屋，其相同房屋與輻射屋距離越遠，與房價之間呈現正向關係，代表房屋邊際價格會越高，另外，在模型(一)中可發現，非線性(Box-Cox)之邊際價格相對較低，而模型(二)中，非線性(Box-Cox)所計算出的邊際價格相較於半對數線性模型之價格較低，但在相距600~800公尺之後，價格受輻射屋影響的程度有逐漸降低。
Abstract
Tests conducted by the Atomic Energy Council in August 1992 revealed that 1,659 houses in Taiwan are contaminated with radiation. Apart from negatively affecting both the wellbeing and the financial circumstances of residents, these houses may also impact on neighboring houses. This study explored whether radiation-contaminated houses affect the prices of neighboring houses, and if so the extent to which distance from the radioactive house in question affects housing prices. We sourced transaction data from real estate market reports on Taipei City and New Taipei City during 1993-1996. Using geographic information systems (GIS), we calculated the distance of specific houses from radiation-contaminated properties, then employed the hedonic price method to study price differences.
Using the semi-logarithmic linear and nonlinear (Box-Cox) hedonic pricing methods, we developed models for gradient distance intervals and straight-line distance intervals. Houses in Taipei City situated more than 1,000 m from radiation-contaminated homes and houses in New Taipei City located more than 800 m apart were eliminated as samples. The conclusions are as follows:
Although Model 1 (gradient distance interval) in Taipei had comparatively lower goodness of fit (39.67%), all other semi-logarithmic linear models showed model fitness of approximately 60% or higher, indicating relatively high explanatory capacity for housing prices. The Box-Cox transformation coefficient θ was below 1% for empirical results in both Taipei City and New Taipei city, which is statistically significantly different from 0. Therefore, the two types of model both have relatively high explanatory capacity.
The hedonic pricing model for housing produced results that were in line with our expectations. The floor area and type of house, specifically townhouse or villa, were shown to be significantly positively correlated with house price. Houses that had been listed on the market for a considerable length of time and properties in apartment complexes were shown to fetch lower prices. All other independent housing variables have a significant influence on house price.
Based on studying houses in Taipei City and New Taipei City situated less than 10 m from radiation-contaminated houses, a positive relationship was demonstrated between increased distance from the radioactive property and housing price, indicating higher marginal price. Model 1 shows that the marginal price calculated by the non-linear (Box-Cox) model is relatively lower, while in Model 2 the marginal price of the non-linear (Box-Cox) model is lower than that of the semi-logarithmic linear model. Once distance was increased to 600-800 m, however, the influence of radiation-contaminated homes on real estate prices was gradually reduced. |
