國外住宅租金普遍具有復歸性質，理性投資者應考慮此性質，於市場高峰期預期較低的未來租金成長率，而以較高資本化率進行評價，於市場谷底期預期較高的未來租金成長率，以較低的資本化率進行評價。故理性預期將導致市場波動穩定。近年相關研究發現，英國投資者對未來租金成長的預期相對較美國及澳洲的投資者理性，投資者理性程度可能具地區差異。本研究利用向量自我迴歸模型驗證台北地區住宅租金具有復歸性質。利用追蹤資料迴歸分析實證台北地區住宅資本化率資料，探討投資者是否具理性預期；研究發現投資者的評價未考慮租金的賦歸性質，不具理性預期，於市場租金低迷時期過於悲觀，於市場租金快速成長時期過於樂觀。此外，本研究探討投資者理性預期的地區差異，發現市場資訊流通較高的台北市，投資者預期相對較台北縣合理，其市場資本化率波動亦較穩定。顯示投資者理性程度對於住宅市場的穩定性具有重要意義。 / Housing rents are widely considered to be mean or trend reverting overseas. Rational investors should consider the reverting potential of rents so that they would expect lower / higher future rental growth rates at rental cyclic peaks / troughs, hence higher / lower capitalization rates. Investors with rationality could appropriately value their housing property, hence they help stability of housing market. Recent studies have found more rationality in the expectations of rental growth of English investors relative to the U.S. and Australian investors. In this study, we use a vector autoregression model to examine the reverting nature of housing rents in Taipei. We use a panel data regression analysis to explore the rationality of housing investors by examining the relationship between current rent level and capitalization rates. The empirical results suggest that investors were too pessimistic / optimistic while rent level is relatively low / high, they have not built the reverting nature of rents into their valuations / or their capitalization rates. Further, we have found more irrationality in Taipei County than that in Taipei City, as a result of the information sufficiency in Taipei City. Hence the volatility of capitalization rates in Taipei County is greater than in Taipei City, indicating that investor rationality plays an important role in housing market.

收益資本化率是不動產估價收益法中最為關鍵的因素，不動產淨收益透過收益資本化率轉換為價值，其步驟看似簡易，卻深藏「失之毫釐，差之千里」的陷井。在國內，商用不動產廣泛運用收益法來估計不動產價值，但收益資本化率之研究卻相當缺乏，相關文獻僅限於國外，因此，本研究以台北市商用不動產為主要研究對象，除了探討收益資本化率之影響因素與實務界在決定過程上的缺失外，並提出較具有解釋力的計算與預測方法。 本研究蒐集自1992年至2008年止台北市商用辦公室之租金與房價資料，利用特徵價格模型將轉換後的租金與房以直接資本化法計算收益資本化率，與問卷結果中可得之影響因素資料進行相關分析、多變量自我迴歸分析，找出關鍵之影響因素，並利用指數平滑法、Box-Jenkins、灰色與VAR模型等時間預測方法來進行收益資本化率之預測。 本研究實證結果顯示，以特徵價格模型計算台北市商用辦公室之收益資本化率，其模型解釋力相當高，並獲致區位上的差異結果。此外，影響因素分為四個面向包含：社會經濟條件、其他投資工具、市場條件、建築物條件等四項，影響因素中除了收益資本化率自身前期為一個重要指標，社會經濟條件之經濟成長率、建築物條件之台電用電不足底度戶數與市場條件之電力(企業)總用電量(十億度)為重要領先指標，其他投資工具之股價指數與社會經濟條件之失業率為同步對收益資本化率有影響性，代表市場條件之不動產景氣領先綜合指標受到收益資本化率之影響而為落後指標，而收益資本化率本身並具有平均數復歸的現象。在未來趨勢的預測方面，單一時間數列適合長期資料之預測模型Box-J進行預測之解釋力（adjR-square）較高，各模型之預測結果平均來說，以Box-J與VAR-I為最有可能之預測值，而最樂觀為VAR-II之預測值，最悲觀為灰色之預測值。 透過國內不動產估價實例進行驗證，本研究結果可以準確進行事後驗證與事前預測，對於提升不動產估價師與投資者收益資本化率的估計與解釋能力上應有所助益。 / Capitalization rate is the most crucial factor in the income approach real estate appraisal method. When this appraisal method is applied, real estate net income is converted into value via capitalization rate. The conversion process is simple and straight forward, but there seems a trap of misuse in real practice. Domestically, income approach real estate appraisal method has been put to use extensively to estimate commercial real estate value. However, research in capitalization rate is in shortage considerably. Related literature merely re-stricted in overseas resource, Hence, our study focus on Taipei office as the main subject to investigate the influential factors of capitalization rate and its shortcoming of real practice, in additional, we propose more concrete explana-tory calculation and estimation methods. Our study collects information on Taipei City office rental and transaction price between 1992 and 2008. Using Hedonic Price Model, the converted rents and transaction price are applied by direct Capitalization Approach to calculate capitalization rate. Thereafter, we compared with the influential factors derived from the survey through correlation analysis, and Vector Autoregression Model (VAR) to find the pivotal influence factors. Meanwhile, we use Exponential Smoothing(ES)、Box-Jenkins、Grey Forecasting Model and VAR time-series estimate method to conduct capitalization rate prediction. The result of this study provides actual proof that to use Hedonic Price Model to calculate office capitalization rate of Taipei, the results are rather ac-curate and indicate a geographical difference. In terms of influential determi-nates, the capitalization rate in itself is an important indicator in the beginning. The Economical Growth Rate and the household under electricity base degree utility of Taiwan Power Company are also significant leading indicators. Stock Price Index, Unemployment Rate and the capitalization rate affect among them. Real Estate Cycle Leading Indicators are affected by capitalization rate and hence a lagged index. Capitalization itself exhibits mean return phenomenon. The future trend prediction of capitalization rate, Box-J and VAR-I give the most likely estimates; VAR-II gives the most optimistic value, whilst Grey Forecasting Model gives the most pessimistic estimation. Through Case study of domestic real estate appraisal to conduct experi-mental verification indicates that the result of our study can accurately carry out aftermath verification and prior estimation. As such, hopefully our empiri-cal study might be able to benefit real estate appraiser and investor to enhance their ability to determine the capitalization rate.