由於不動產投資信託(Real Estate Investment Trust, REIT)的投資門檻低,又同時具備股票市場高報酬和不動產市場抗通膨等特性。近年來已成為熱門的投資商品,投資風潮延燒全球。 台經院研究員預期於2013年下半年開始房市投資氛圍轉差,最快2015年資產泡沫化的疑慮可能浮現。因此,必須審慎思考如何降低投資風險,且能準確預測不動產投資信託的價格,不僅能提供投資人在選擇投資項目時,能有較準確的依據;更能避免受到房市泡沫化影響。而預測對投資商品而言,將扮演著極重要的角色。本研究結合總體經濟變數和影響股市行情相關變數來探討不動產投資信託過去績效,將研究範圍於台灣不動產投資信託,運用灰色關聯分析與類神經網路,以期能找出相關影響因子對不動產投資信託股價之預測。若能預測出不動產投資信託價格之走勢,則投資人可提高對投資不動產投資信託的意願。 本研究運用灰色關聯分析結果發現,REIT 分別受重貼現率、原油、黃金、短期融通率、建築貸款餘額、國民平均所得、台灣出口貿易總額、道瓊指數及其本身過去績效影響最深。再利用類神經網路驗證灰色關聯分析效果,有效的應用在許多預測的問題,主因為大部份之預測問題都屬非線性模式,而倒傳遞類神經網路具備建構非線性模型之能力。本篇是利用灰關聯分析在眾多因子中篩選出關聯度較高之因子,再將此類因子置入智慧型預測模式中進行訓練與預測,運用此模式建立較高準確度之預測模型,幫助投資人有效預測不動產投資信託之價格。
In recent years, Real Estate Investment Trust (REIT) has become a popular investment around the world. REIT has the advantages of inflation-resistance as well as real estate and high return on stock market. Researchers of Taiwan institute Economic Research expect that the situation of investment in real estate will become worse in the second half of 2013 and the Burst of Real Estate Bubbles could happen around 2015. Therefore, by considering how to reduce the risk of investing real estate and anticipating return rate correctly, those techniques not only can help the investors choose their investment but also keep them from the effect of the Burst of Real Estate Bubbles. This research combines the factors of acroeconomics and the performance of investing real estate in the past focuses on real estate in Taiwan by using Grey Relational Analysis and Artificial Neural Network. Therefore, the main purpose of prediction is to estimate the oncoming events or situations in advance and provide the best information to management level to detect those uncertain conditions and help reduce the risks during decision-making process. After that, apply the result on production output assessment to increase the accuracy of forecast and provide the consultation to related industries for production forecasting.