This study examines whether nonlinear adjustment of short-term deviations impacts US real estate market returns by applying an exponential smooth transition threshold error-correction model with Generalized Auto Regressive Conditional Heteroscedasticity (GARCH) (ESTECM-GARCH). Empirical results demonstrate that the ESTECM-GARCH captures the dynamics of returns more effectively than the Error-Correction Model (ECM) and Exponential Smooth Transition Error-Correction Model (ESTECM). Consequently, the nonlinear behaviour of returns is driven by momentum noise traders and heterogeneous arbitrageurs in Real Estate Investment Trust (REIT) markets.