Forecasting the REITs' returns in US: New evidence from structural time series model

The purpose of this paper is to apply the structure time series model to forecast the REITs' returns in US by using monthly data from January 1994 to November 2009. The primary advantage of this model is not only taking into account trends, seasonal factors and cycle effect but also that reflects the realistic situation of REITs, therefore, to bring about more accurate short-term forecasting. Empirical results indicate that cycle is much more explainable to the returns of industrial REIT and manufactured home REIT because industrial REIT and manufactured home REIT are lower and stable returns than other REITs. And the impact of shock occurred in September 2008, the industrial REIT and manufactured home REIT's returns are slowly decreasing but stable. Furthermore, in accordance with the model forecasting results, a directional path was precise forecast from March 2009 to November 2009 period belonging to in-sample forecast. And we also obtain out-of-sample industrial REIT and manufactured home REIT's returns in 2010, they are decreasing gradually but stable and low risk relatively. This important information can provide investors with the best investment choice when investors are in the face of economic shock.