Rational Expectations in Housing Markets: The Case of Survey Forecasts

Authors: Svetlana Doronkina, James Forest

SUNY Campus: SUNY New Paltz

Presentation Type: Oral

Location: UUW 325

Presentation #: 9

Timeslot: Session A 9:00-10:00 AM

Abstract: This paper examines the rational expectations hypothesis within the U.S. housing market by evaluating forecast efficiency and bias through econometric methods. The study focuses on five key housing indicators—building permits, construction spending, existing home sales, new home sales, and housing starts—across both pre- and post-2006 housing boom periods. By employing Augmented Dickey-Fuller (ADF) tests, Engle-Yoo (EY) and Dynamic Ordinary Least Squares (DOLS) cointegration regressions, Mincer-Zarnowitz (MZ) bias tests, and anchoring bias tests, the research uncovers varying degrees of stationarity and cointegration across indicators, revealing inefficiencies in forecast accuracy. Results indicate that while some indicators exhibit strong long-term equilibrium relationships between forecasts and actuals, others, such as construction spending, display a significant bias and low predictive accuracy. Housing starts and existing home sales reveal structural breaks that impact forecast efficiency, and anchoring bias is significant in certain series, reflecting cognitive distortions in housing market expectations. These findings challenge the assumption that housing market expectations fully incorporate available information and contribute to the broader literature on economic forecasting by identifying the limitations of rational expectations in the housing market. This work may be especially valuable for policymakers and economists seeking to enhance forecasting methodologies, especially in markets prone to volatility and speculative bubbles.