Two studies of Japan-REIT performance : modeling risk and tracking property-level performance

Abstract

This paper is intended to recognize the performance of REITs in Japan (J-REITs) by conducting two kinds of studies in a REIT-level and an underlying property-level: first, to do "factor loadings" that identify systematic risks of long run investment performance in J-REITs; second, to demonstrate "Pure Play Indices," segment-specific indices of REIT-based property market returns by tracking monthly REIT return data and property holding data. The first study employs the Fama-French three-factor model for monthly J-REIT returns from September 2001 to September 2008. This investigation upgrades past similar research with longer data periods in a two-stage regression (a time-series regression and a cross-sectional regression) for all the listed J-REITs. Nevertheless, the model results in a limited explanatory power for the J-REIT performance, probably due to too short a market history, as in the past research. The second study applies the Pure Play Indices, originally proposed by Geltner and Kluger [1995, 1998], to the J-REITs for office, residential, and retail segments since January 2006 when the J-REIT market became sizable enough for study. The developed Pure Play Indices perform similarly with the J-REIT return indices, except the Pure Play Residential Index during the down market due to the effect of non-target segments within the J-REITs. The reason for this effect will require a further study. As the market matures with more data accumulated, this two-fold study that shows demonstration of returns from J-REITs will become more valuable to derive risk of J-REITs and different types of information of properties.