Navigating Japan's Real Estate Market: A Predictive Framework for REIT Index Returns Using Macroeconomic and Sentiment Data

This study develops a robust time series model to forecast Japan's REIT returns by integrating macroeconomic and sentiment indicators. Existing research focuses on traditional variables like interest rates and inflation but neglects sentiment factors and time series dynamics. The study uses monthly TOPIX REIT returns (2004-2023) as the dependent variable, incorporating predictors such as bond yields, CPI growth, exchange rates, consumer confidence, and volatility indices. Through Newey-West adjusted OLS regression, the model accounts for autocorrelation and heteroskedasticity, demonstrating strong predictive power. Results show significant impacts from bond yields (=0.265, p<0.01) and exchange rates (=-1.341, p<0.01), while market panic (=-0.157, p<0.01) negatively correlates with returns. The model achieves an adjusted R of 0.47 in-sample and superior out-of-sample performance (RMSE=0.127), outperforming historical benchmarks. It validates the value of sentiment indicators in REIT forecasting and provides a dynamic decision-making framework, especially during market turbulence like the 2020 pandemic. The study advances small-sample financial time series modeling and highlights the critical role of behavioral factors in REIT market dynamics.