空間特徵分量Durbin迴歸評估都市多元開放空間價值──以桃園市與臺中市為例

Evaluation of Diverse Open Space in Urban Area by Spatial Hedonic Quantile Durbin Regression: Case Studies in Taoyuan City and Taichung City

蔡孟珂(Meng-Ke Tsai)

指導教授 ： 吳珮瑛

國立臺灣大學/生物資源暨農學院/農業經濟學研究所/碩士(2018年)

https://doi.org/10.6342/NTU201802770

摘要

都市的開放空間提供環境、社會與經濟各面向的效益，帶給居民良好的生活品質，是都市永續發展之重要基石。然而，當人口壓力造成都市擴張，而將開放空間挪作他用時，會導致開放空間數量、面積減少、分散破碎、種類單一化，使開放空間的各項功能退化，居民也失去原有開放空間帶來的效益。因此，需適當評估開放空間價值，才能掌握當開放空間的數量與品質變化時居民享有之效益的改變。故本研究選擇國內人口成長最快速之桃園市與臺中市之都市計畫區內的開放空間，由中歸納整併出兩個城市共同具有的「鄰里公園」、「都會公園」、「緩衝綠地」、「廣場」及「農地」五種開放空間，同時參酌地方景觀特色，桃園市另外納入「埤圳」之開放空間。
　　由於開放空間帶給居民多方面的有形無形效益，這些可以反映在各類型住宅房屋市場價值的高低，因此可採特徵價格法評估各類型開放空間帶給居民的效益。然房價及影響房價高低之各項因素，經常涉及空間自相關問題，空間Durbin模型可以解決相關的問題；此外，解釋房價的諸多因素對不同房價的影響，一般而言並不相同，此種情況則需採分量迴歸，方能分析極端價位下開放空間的效益。過去文獻並未有同時考量上述問題的估計模型，本研究則首度將兩方法結合，使用「空間特徵分量Durbin模型」分析實價登錄資料上由2012年8月至2017年7月止，桃園市97,244筆、臺中市66,107筆之住宅相關資料，評估住宅周邊之多元開放空間，在不同距離、面積、空間分散程度與空間種類等不同空間數量及品質組合下對居民造成的效益或損失。
　　評估結果顯示，桃園市除農地與埤圳之外，各類空間離住宅距離越近則邊際效益越高，至於臺中市除緩衝綠地之外，各類空間離住宅距離越近則邊際效益越高。其中，兩個都市之鄰里公園離房屋越近對房價最高之95分量房屋的邊際效益越高，而房屋附近開放空間越分散對房價最高之95分量房屋的邊際效益及邊際損失最大。但開放空間的分散程度增加可能因為原有開放空間遭破壞而變得破碎，導致居民的損失；也可能是原有開放空間分布不均且數量少，分散時開放空間數量反而增加且分布均勻，帶給居民效益。在開放空間組合方面，桃園市50分量房價之房屋與房價最高95分量之房屋2公里內有6種類型空間時總效益最大，而最低5分量房價房屋總效益的最高組合為「都會公園──緩衝綠地──廣場──農地」；對臺中市最低5分量房價之房屋總效益最高組合為「鄰里公園──緩衝綠地──廣場──農地」，也是由4種類型開放空間組成；顯示開放空間的多元性有價值。

關鍵字

開放空間 ； 空間特徵分量Durbin模型 ； 都市永續發展 ； 空間離散指標 ； 都市擴張 ； 地理資訊系統

並列摘要

Open space in urban areas provides residents with benefits in environmental, social, and economic aspects, which forms the foundation of the sustainable development of urban areas. However, as population growth causes urban sprawl, the changes of open space for other purposes result in the area reduction of open spaces, decrease in the number of open spaces, the fragmentation and simplex of open spaces. The residents will then lose their benefits from all types of functions provided by open spaces. As a result, a proper evaluation of benefits for open space is necessary to understand how and in what way that residents’ benefits might vary from the change of quantity and quality of open spaces. Thus, this study demonstrates such ideas for Taoyuan City and Taichung City in which there is rapid population growth among six municipalities in Taiwan. Open spaces in these two cities can be categorized as “neighborhood park,” “metropolitan park,” “greenbelt,” “square,” and “farmland.” In addition, “reservoir and canal wetland” is a unique landscape in Taoyuan City, so it is included as one additional open space in Taoyuan City.
Hedonic price method can properly evaluate the tangible and intangible benefits that open spaces bring to urban residents. Housing prices and factors influencing housing prices usually involve spatial autocorrelation. Spatial Durbin model can solve such problem. Furthermore, a specific factor normally has different impacts on housing prices. Quantile regression model enables the assessment of open space in specific housing price percentiles and thus is able to achieve the purpose. There is no study in the past which takes both issues into account. This study is the first research that combines two methods to solve the issues of spatial autocorrelation both in the housing price variable and in its corresponding explanatory variables, and analyze the performance for specific housing price percentiles. A model of spatial hedonic quantile Durbin regression model is developed by this study. Two sets of housing data for Taoyuan City and Taichung City from August 1 in 2012 to July 31 in 2017 including 97,244 house transactions and 66,107 house transactions respectively are analyzed for such purpose. The diversification of open space is represented by its distance and corresponding to size to each transacted housing, and a spatial dispersion index (SDI). The change of benefits evaluation is accomplished by different aspect combinations of open spaces mentioned above.
The results show that for Taoyuan City, except “farmland” and “reservoir and canal wetland,” the closer of a specific transacted housing to all other four types of open spaces is the larger the marginal benefit is. In contrast, for Taichung City, except “greenbelt” the closer of a specific transacted housing to all kinds of open space is the lower the marginal benefit is. Among these, “neighborhood park” has the largest marginal benefit for the 95th percentile housing prices in both cities. Similarly, the marginal benefit or marginal loss is the largest with larger SDI for the 95th percentile housing price in both cities.
Furthermore, the results also reveal the benefit of open space diversity. The highest total benefit in Taoyuan City for both the 95th and 50th percentile housing prices are the composition of open space in the 2 kilometers buffer zone with 6 kinds of open spaces. For the 5th percentile housing price, the highest total benefit is for the composition of open space with 4 kinds of open spaces. The highest total benefit for Taichung City is also composed of 4 kinds of open spaces. In brief, the residents prefer the composition of 4 kinds and more than 4 kinds of open spaces. Besides, the increase of SDI for open space might be arisen by the destruction and/or expansion and disperse. The former leads to smaller size of open space. The latter means the number of open space increases but unevenly distributed in the buffer zone. The benefits provided to residents could increase or decrease one way or the other.

並列關鍵字

Open Space ； Spatial Hedonic Quantile Durbin Regression Model ； Sustainable Urban Development ； Spatial Dispersion Index ； Urban Sprawl ； Geographic Information System

參考文獻

『公司法』，2015。（http://law.moj.gov.tw/LawClass/LawAll.aspx?PCode=J0080001）（2018/3/6）

『自然』雜誌社論，2008。「全球都市化的挑戰」，『知識通訊評論』。72期，14-15。

『桃園市市立國民中小學實施學校規模總量管制作業要點』，2016。（http://www.tcjhs.tyc.edu.tw/school/data/law/201605110942440.pdf）（2018/2/15）

『桃園市市立國民中小學學區劃分原則』，2016。（http://163.30.174.3/xoops//uploads/tadnews/file/20160512_1.pdf）（2018/2/15）

『桃園縣埤塘水圳保存及獎勵新生利用自治條例』，2009。（http://law.tycg.gov.tw/LawContent.aspx?id=FL051054）（2018/4/22）