特徵價格模型是傳統常被使用於不動產大量估價的模型，由於模型將造成所有價位的不動產其特徵都具有同樣的邊際價格而無法解釋現實不動產特徵的各種可能狀況，故引發本研究利用分量迴歸建立大量估價模型之動機。研究利用台灣不動產成交行情公報的資料進行台北市大廈的實證分析，針對特徵價格法的延伸與估價準確度做檢視。嘗試應用分量迴歸建立大量估價模型，討論住宅特徵對於價格的邊際影響力於不同價位的住宅是否存在差異，並討論分量迴歸模型的估價精確度。研究採用交互驗證法與重複實驗30次討論模型的估計效果，並利用平均絕對百分比誤差（MAPE）以及命中率（Hit Rate）做為模型預測優劣程度的衡量標準，以討論分量迴歸模型是否可以較最小平方特徵價格模型有更為準確的估計表現。實證首先探討價格分量之下各住宅屬性對於價格的影響狀況，得到大部分住宅特徵對於價格的邊際影響力的確會因住宅價位的不同而有所差異。在估價準確度的部份，經測試得到利用分量迴歸建立大量估價模型的估價效果達研究的預期目標，且其估計表現優於最小平方特徵價格模型。 / 藉由分量迴歸模型，得到隨著住宅價位的增加，坪數與屋齡對於價格的影響力並非呈現一致的趨勢；坪數輪廓與屋齡輪廓出現轉折也為變數增加二次項變數的原因得到實證依據。重複實驗30次的整體表現，分量迴歸模型的MAPE較最小平方迴歸模型低了1.687％；誤差落在正負10％的Hit Rate較最小平方迴歸模型高了3.81％；誤差落在正負20％的Hit Rate較最小平方迴歸模型高了5.14％。30次的實證為分量迴歸模型的估價表現更優於最小平方迴歸模型得到較具說服力的結果。 / Hedonic pricing models are traditionally used for real estate automated valuation models. Because the conditional mean calculated by OLS does not give a complete description of the relationship between dependent variable and independent variables, which leads to the motive of this study. This study inspects the extension of hedonic pricing models and appraisal accuracy, and we attempt to apply quantile regression to real estate automated valuation models and discuss the difference of the marginal contribution in each individual characteristic under different price level. Our study adopts cross validation and repeats empirical process for 30 times, and we use MAPE and hit rate to evaluate accuracy and argue if quantile regression models have better estimation. The empirical results show that the marginal contribution of housing area and age changes with price level; the turning points of area curve and age curve show empirical evidence for including square variables. The entirety performance of repeated experiments points out that the MAPE of quantile regression model is 1.687% lower than OLS model; as error ranged between 10% to -10%, the hit rate of quantile regression model is 3.81% higher than OLS model; as error ranged between 20% to -20%, the hit rate of quantile regression model is 5.14% higher than OLS model. The 30 times experiment of quantile regression models shows a much more persuasive result than OLS models.

從1960年代開始，公部門基於稅務處理需求，使得電腦輔助大量估價（Computer Assisted Mass Assessment，CAMA）成為輔助的工具，大幅提升了估價的效率。在1990年代，金融機構因不動產證券化的發展及不良資產估價等業務，而衍生了對大量不動產進行估價的需求，同時在電腦與統計模型的進步之下，自動估價模型（Automated Valuation Models，AVM）應然而生，並被廣泛應用在金融市場。由此可知因為不動產經濟活動的熱絡發展，大量估價的需求日益增加，其具備的客觀與效率等優點更彰顯其重要性。 雖然大量估價的需求日益增加，然而過去對於估價準確性相關研究，主要著重在估價理論與技術層面、估價行為對估價結果的影響、探討個別估價和大量估價的估值比較，而較少單獨探究影響大量估價準確性的因素。由於特徵價格理論隱含不動產高度異質的特性，不動產價格受到總體經濟、政策、住宅屬性、公共設施、區位等因素影響，然而前述因素是否會對估價準確性造成影響?造成影響的因素為何?為本文所欲探討之問題。 本文在實證部分分成兩階段，首先以特徵價格理論為基礎，利用實價登錄資料建立大量估價模型，以MAPE與Hit Rate來衡量估價準確性，結果指出MAPE達到14.19%，而正負誤差10%的命中率為47.18%、正負誤差20%的命中率為74.75%，跟過往研究所建立的大量估價模型相比具有相當的水準，顯示出官方性質的交易資料具有一定的可信度。在建置大量估價模型後，本文以模型價格及成交價格間的比值作為劃分估價準確程度的依據，以多項羅吉特模型進行實證分析，結果指出住宅大樓、捷運站周遭住宅、大坪數住宅估價結果容易呈現低估情形；而新市區中心估價結果容易呈現高估的情形；另外比較特別的是舊市區中心、北郊區估價結果較容易呈現高估及低估，換言之在這兩個區域估價容易得到不準確的結果。 / Since 1960s, public sector began to take advantage of computer assisted mass assessment(CAMA) based on taxation services and greatly improved the efficiency of appraisal. In 1990s, financial institutions due to the development of securitization of real estate and non-performing asset valuation and other services, generating the demand of mass appraisal. Simultaneously, due to the development of computer and statistical models gradually progress, bring in automated valuation models(AVM) in the financial markets. Hence, with the real estate economic activities gradually booming, the increasing demand for mass appraisal, which has the objective of efficiency and other advantages will be more to highlight its importance. While the increasing demand for mass appraisal, but past studies about the accuracy of appraisal, mainly focused on the theoretical and technical aspects, the impact of behavior on the valuation results, and to explore appraisers and mass appraisal of the valuation. However, past studies less focused on a large number of factors affect the accuracy of the appraisal. Since the hedonic price theory implies highly heterogeneous characteristics of real estate, real estate prices affected by factors of macroeconomic, tax policy, housing properties, public facilities, location and so on, but whether the aforementioned factors will affect the valuation accuracy?Is this research seeking to explore the issue. In this paper, the empirical section is divided into two stages, first with the hedonic price theory based on the use actual price registration to establish the mass appraisal models, and base on MAPE and Hit Rate to measure the accuracy of the appraisal, the results indicate MAPE reached 14.19%, while the margin of error of 10% hit rate of 47.18%, 20% hit rate is 74.75%. Compared with the past studies, this model has established a great performance. This research proved that the official data with reliability. After establishing the mass appraisal models, the research use model prices and the transaction price ratio as the basis for division between the accuracy of the appraisal and use multinomial logistic model to conduct empirical analysis. The results indicated that the residential building, housing around MRT stations, the big area housing was prone to result underestimate valuations, the new urban center appraisal results likely to show overvalued valuations. On the other hand, old city center and the northern suburbs results presented overestimate and underestimate valuations simultaneously, in other words, that is usually get inaccurate results in these two regions.

長期以來，不動產估價相關研究皆著重在估價理論與技術的改進，以使最後估值能更為精確。然而實際上，估價人員的行為才是影響價格評定的本質因素；尤其國內、外對於市場比較法的相關研究，皆發現估價人員有偏離理論規範的現象。故本文第一部份是分析估價人員系統行為偏誤問題；第二部分探討特徵價格模型的偏誤，並應用分量迴歸改進不同高低價格的估價模型準確度；第三部分比較估價師個別估價與大量估價模型之偏誤與差異；為提升客觀估價方法，第四部分探討以估價師估值樣本產生最可能價格與機率的方式。本研究藉由估價師估價行為與大量估價模型不同面向的研究，重新檢視市場比較法的各種偏誤現象。 一、市場比較法估價之系統行為偏誤 市場比較法是估價人員最常使用的方法，但在估價過程中主觀判斷部分相對較多也最易引起爭議。過去市場比較法研究著重在估價理論與技術改進，以減少主觀判斷造成的隨機偏誤，然而，研究指出行為偏誤造成的系統偏誤，其持續性的偏誤對於估價偏誤有重要影響。過去探討估價行為的研究，受限於資料樣本取得，幾乎都是以問卷或實驗得到結果，而非實證結果。本文則是以實際估價報告書進行實證分析，以鄒檢定(Chow Test)及似無相關檢定(Seemingly Unrelated Regression)，發現勘估標的估值與比較標的市場成交值兩模型係數有顯著差異，表示估價人員估值模型起始點高於市場成交值模型，估價人員對於使用類型有高估現象，對於面積估計有低估現象，支持估價人員有系統行為偏誤。此外，也發現評估不同市場時，行為偏誤程度亦不同，估價人員於評估拍賣價格時，因估價經驗與比較案例較少，較傾向參考比較案例，行為偏誤亦較小，僅面積係數有差異；但於評估正常市場價格時，估價行為偏誤較大，區位、使用類型及面積係數皆有差異。 二、不同高低價格不動產估價模型之偏誤改進－分量迴歸模型之應用 隨著國內不動產市場M型化推案趨勢，非典型住宅（如：高總價豪宅和低總價小套房）類型逐漸增多，對於此類型產品的估價精準度也需要提升。從過去研究發現，最小平方迴歸估計忽略各特徵屬性對價格條件分配的差異。本研究乃以分量迴歸方法建立住宅大量估價模型，藉以瞭解住宅特徵對於不同價格分量的差異，實證結果發現以最小平方迴歸模型估計相較於分量迴歸，對於一樓、頂樓、車位、區位等變數有高估或低估的情形。比較估值模型預測精確度，本文透過30次重複實驗，發現分量迴歸對於兩側尾端樣本有較佳的預測能力。從實證方法而言，本文改進以最小平方迴歸模型對兩尾端價格高估或低估問題；就實務應用方面，隨著不動產產品差異度增加，以及新版巴塞爾協定（Basel Ⅱ）實施對不動產價值更新的需求，分量迴歸模型可提升兩尾端估計精確度，並提供住宅大量估價系統另一種資產重估方法。 三、大量估價與個別估價之差異分析 不動產估價研究可分為個別估價與大量估價兩類，過去有關個別估價研究多為估價行為或估價方法改進，而大量估價研究則多是運用數量方法建立模型，由於兩者建構方式不同，欲同時進行兩者實證比較並不容易。本文藉由相同勘估標的，取得估價師個別估價資料，與資料庫建立的自動估價系統，進行案例選取、權重調整及最後估值三階段的比較分析。經本文差異比較發現，前二個階段的比較結果，皆是自動估價系統優於個別估價，從估價行為觀點，顯示自動估價系統具有較客觀且符合估價理論程序優點。然而第三階段的估值比較，發現自動估價系統的表現並不如預期，此結果與過去研究結論不同，分析兩者差異值大的案例，發現資料庫樣本較少地區以及非典型住宅類型（面積過大或過小），是造成個別估價估值與自動估價系統估值差異大的原因，顯示自動估價系統有其適用限制，未來若增加資料庫樣本或次市場模型，兩者估值差異將可獲得改善。 四、不動產估值認知與估值機率 本章以估價師估值資料為樣本，藉由蒙地卡羅模擬估值機率模型，其模擬結果可客觀估計最可能價格（the most probable price）與機率。由於資料樣本來自於專業估價師的判斷，估值分配代表的是估價師所估計的可能估值結果；相較於過去僅以單一估值決定資產價值，藉由多位專業估價師所判斷結果而建立的估值區間與最可能價格及機率，應該是更具客觀性且接近市場價值，而且估值機率的呈現將有助於估價報告書使用者評估存在於資產價值的上方（upside risk）及下方風險（downside risk）。 / In the real estate appraisal research, appraisers and mass appraisal are the two methods most often used in the sales comparison approach. Past studies focused on appraisers behavior and mass appraisal model, lack of compared the difference by appraisers for the same objects as well as mass. The first essay reviews the behavior literature regarding real estate appraisers and summarizes the two hypotheses for departure from normative models. The study is based on appraisers’ reports analyzes for the appraisal valuation and the market transaction by hedonic price models. Among the 112 appraisal reports of the transaction data of 230 comparables and the valuations of 224 subjects, the results reveal the effect of the variables on the appraisal valuation are not consistent with those of the market transaction. In addition, comparing the appraisers’ behavior on the general market with the auction market, the result found is the differences between the valuation models and the transaction models are less than the models on the auction market. The empirical evidences support the two hypotheses and can be explained plausibly by the appraisers’ behavioral contention. The second essay analize the current domestic trend of residential types, it is easy to find that high-priced dwelling units and low-priced dwelling units keep gaining popularity. Thus, the estimation of these two types of residences should be more precise. Since ordinary least square regression can not signify the variation caused by different quantile functions of a conditional distribution, this study estimates the housing price by quantile regression. We compare the models by using ordinary least square regression and quantile regression. The empirical results show that the distributions of some variables, such as first floor, top floor, parking lot, location, are different between two models. These are easily to be underestimated or overestimated when ordinary least square regression is applied. Based on thirty repeated experiments using random sampling, the results of hit rate and mean absolute percentage error show that quantile regression estimates more accurately on two-tailed distribution. For mass appraisal application, a quantile regression advances the estimate on two-tailed price and provides a new method on assets reevaluation of banks.

如何客觀有系統的估計公告土地現值一直是土地估價研究領域的熱 門話題，目前公告土地現值的查估，多數以區段地價作為宗地地價，受限 人為主觀與人力不足的缺點，查估的結果並不一定能有效反應各區段間地 價差異。由於以往房地產實證研究的領域中，特徵方程式一直是受到廣泛 運用的工具，然多數著重於各別宗地價格進行模型設計，較少以地價區段 範圍建立估價模型。本研究以三峽鎮住宅區民國89 至98 年區段地價進行 實證分析。變數的選取主要是配合「地價調查估計規則」所規範影響普通 住宅用地區域因素基準，並將全部變項納入複迴歸模型中，先測試綜合影 響程度，再將未符合預期及篩選合理顯著變項重新建立區段地價估價模 型。實證顯示接近公車站牌之程度、區段內道路規劃及開闢建程度、景觀 有無、保排水良否、地勢是否高低起伏、至國中小距離、至市場超市距離、 至三峽老街距離、停車是否便利、至墓地殯儀館火葬場距離及是否具發展 潛力等11 項變數達顯著水準，於20%內之Hit Rate 達91.18%，MAPE 亦僅 7.9%，均能符合預期表現。本文透過區段地價估價模型之建立，提供電腦 輔助區段地價估價可行方案，藉以增進公告土地現值評估客觀及科學化程 度。 / How to estimate the announced current land value objectively and systematically is always a hot issue in land valuation research field. And, since the announced current land value is the foundation for levying the land value increment tax and compensation when land expropriation, the risk of unfairness might happen if the announced current land value is not objective and systematical. Under the announced current land value system, most parcel land values are produced using the district land value. Although decades of valuation experience by assessors, the district land value would not necessarily reflect fundamental value effectively. Taking into consideration of the difference between the degree the zoning affect the land value and the heterogeneity characteristic of land, this paper construct district land value model on different zoning. The empirical study region is the residential zoning area in the Sanshia Township, for its landscape with new and old mixed buildings, featuring metropolitan development characteristic, and stable sales transaction volume. The empirical time period is from 2000 to 2009. The district land value estimated from sales, collected from the Shulin Land Office, is the dependent variable. The selection of the independent variables is in line with the region factors of common residential area regulated by “The Regulations on the Land Value Investigation and Estimation” after combining similar attributes for easing the bias possibility from co linearity. The empirical result shows the significant variables are the ratio of constructed road area to total area within the land value district, parking convenience, development potentiality and the distance from bus station, junior, elementary schools, market, service facilities, graveyard, etc. The model fit is good with adj-R2. This paper hopes to increase the automation degree of the announced current land value and make the announced current land value objectively and systematically by establishment of the district land value model.

市場比較法估價過程需要經比較、分析及調整三階段，而估價師於個別估價應用中因缺乏標準化依據，造成估價過程常被質疑過於主觀且偏重經驗法則，導致估價結果產生因人而異現象。基於此本研究乃建立大量估價模型，運用估價師進行市場比較法行為邏輯，在選取比較案例階段採用明科斯基距離概念，並結合特徵價格理論，建立大量樣本的客觀標準化估價模型。藉由此模型分析，估價師不需要主觀預測，可改善過去估價結果不一致情形，並達到大量估價目的。 透過隨機抽樣的重覆實驗，實證結果發現，以模型的平均百分比預測誤差與命中率比較，整體而言未劃分次市場估價模型皆較劃分次市場準確，且其穩定度亦較高，而劃分次市場較不準確的原因，在於舊市區的表現不穩定，與市郊區的估值表現較差所致。經本研究結果，不論劃分次市場與否，標準化大量估價模型大致上均達到研究預設水準，然劃分次市場模型雖有時較準確但未優於未分區模型甚多，顯示適當劃分次市場可提高準確度，但資料太少時，劃分次市場可能產生統計偏誤問題。 / The process of the market comparative approach includes three stages：comparing ,analyzing and adjusting. Real estate appraisers lack the standardized basis in the individual appraisal application, so they are often challenged by subjectivity and stressing experience, which leads to the phenomenon that appraisal results are always different from person to person. Based on this, our research establishes the automated valuation methods. By applying the appraisers’ behavior in market comparative approach, we use minkowski metric in selecting comparative subject, and associate with hedonic price theory to establish objective standardized real estate appraisal model. By using this model, the appraisers can avoid subjective forecasting, reduce inconsistency, and therefore achieve the goal of mass appraisal. Through the repetition experiment of random sampling, we compare MAPE and Hit-rate between models. The result shows that non-delimited markets are more accurate and stable than delimited markets. The reason for this is the instability caused by older district of the city and bad performance in the suburban area. After these findings, no matter delimited market or not, the standardized mass appraisal model achieved the research standard we had set in advance. Sometimes the model of delimited market is more accurate, though not by a significant amount, it shows that higher accuracy can be expected through adequate market segmentation, but will lead to bias when lack of datum.