美國REITs報酬率的預測因子之研究

Abstract

本研究主要探討美國不動產投資信託(REITs)報酬率的預測因子，利用兩種檢定模型：簡單線性迴歸模型與複回歸歸模型檢視每個預測因子的解釋能力，推導出美國REITs報酬率的資產定價模型。本研究共選取五個預測因子，分別為：公司規模、週轉率、帳面價值與市場價值比(B/M)、益本比(E/P)、股利與股價比(D/P)。同時，根據Fama and French(1993)的三因子模型，探討市場風險溢酬(Rm-Rf)因子、規模溢酬因子(SMB)以及淨值市值比溢酬(HML)因子對於REITs報酬的解釋能力。由於美國REITs市場在1990年之後，產生結構上的變化，因此，本文將探討1990年之前以及1990年之後兩個時期，REITs報酬率之預測因子的解釋能力，經由實證分析發現，在1990年之前，只有公司規模因子具有預測力，能夠解釋REITs報酬的變異，而在1990年之後，五個預測因子皆能夠解釋REITs報酬的變異；另外，在三因子模型表現方面，在1990年之前，市場溢酬因子以及HML無法解釋REITs報酬的變異；相反的，在1990年之後，市場溢酬因子與高淨值市值比的投資組合則會有較佳的解釋能力。 最後，本文進行穩健性檢定（robustness test），判斷其是否能夠與複迴歸檢定模型產出一致的結果。實證發現，在穩健性檢定之下，也能夠得到較一致的結論，即兩種不同的檢定統計方法可以產出具有一致的結果。

In this study, we mainly discuss the predictors of the rate of return of American real estate investment trust (REITs). We totally choose five factors as the predictor and use two kinds of models: The simple linear regression model and the multiple linear regression models to examine the explanation ability of each predictor. The five factors are company size, turnover ratio, book to market ratio (BM), earnings to price ratio (EP), and dividend to price ratio (DP). At the same time, according to three factor model of Fama and French (1993), we examine the relation among excess return of REITs and three factors: market risk premium factor (Rm-Rf) , size premium factor (SMB) and book to market ratio premium factor (HML). Because there was a structure change in the REITs market that occurred after 1990, we will discuss the explanation ability of predictors of the two different sub-periods: pre-1990 and post-1990. The determinations of the expected return of REITs differ in the sub-periods. In the pre-1990 sub-period, only size factor predict expected REITs returns. In the post-1990 sub-period, size factor, turnover ratio, BM ratio, EP ratio, and DP ratio predict expected REITs returns. In the three factor model analysis, the market risk premium, and book to market ratio premium (HML) could predict the excess REITs returns in the post-1990 sub-period. The most important concept is that we find that the explanation ability of the predictors is very different in the two sub-periods. More specially, the explanation ability of predictors in the post-1990 is better than in the pre-1990.