Testing for Long Memory in REIT Returns

This study examines the long memory properties of composite, equity, mortgage, and hybrid real estate investment trust (REIT) returns by using semi-parametric and wavelet estimators. In particular, this paper applies the GPH semi-parametric estimator, the Haar and the Daubechies wavelet procedures to investigate the long memory properties of REIT returns. The results from the various procedures reveal that composite, equity, mortgage, and hybrid REIT returns are long memory processes with anti-persistence. The existence of long memory suggests that the dynamics which govern the four return series contain predictable components. This finding indicates that the markets for composite, equity, mortgage, and hybrid REITs are inefficient. The fact that these markets are inefficient suggests that investors can devise profitable strategies by using historical data or past information.