Linear and Nonlinear Predictability of International Securitized Real Estate Returns: A Reality Check

This paper examines short-horizon return predictability of ten largest international securitized real estate markets, with special attention paid to exploring possible nonlinearity-in-mean as well as nonlinearity-in-variance predictability. Although international securitized real estate returns are generally not predictable based on commonly used statistical criteria, there is much evidence for the predictability based on economic criteria (i.e., direction of price changes and trading rule profitability), which is more often due to nonlinearity-in-mean. The forecast combinations for various models appear to improve the forecasting performance, while the allowance of data-snooping bias using White’s Reality Check substantially mitigates spurious out-of-sample forecasting performance and weakens otherwise overwhelmingly strong predictability. Overall, there is robust evidence for the predictability in many international securitized real estate markets.