Improving ex ante accuracy assessment in predicting house price dispersion: evidence from the USA

TL;DR

This paper introduces a novel parametric bootstrap method to improve ex ante prediction accuracy assessment of house price dispersion measures in the USA, especially under unanticipated shocks, enhancing real estate market risk management.

Contribution

It proposes a new approach using a parametric bootstrap under misspecified models to better evaluate prediction errors during unexpected market events.

Findings

The method provides more accurate prediction error estimates during shocks.

Traditional methods underestimate risks associated with unforeseen market changes.

Application to U.S. real estate data demonstrates practical effectiveness.

Abstract

The study focuses on improving the ex ante prediction accuracy assessment in the case of forecasting various house price dispersion measures in the USA. It addresses a critical gap in real estate market forecasting by proposing a novel method for assessing ex ante prediction accuracy under unanticipated shocks. The proposal is based on a parametric bootstrap approach under a misspecified model, allowing for the simulation of future values and estimation of prediction errors in case of unexpected price changes. The study highlights the limitations of the traditional approach that fails to account for unforeseen market events and provides a more in-depth understanding of how prediction accuracy changes under unexpected scenarios. The proposed methods offers valuable insights for real estate market management by enabling more robust risk assessment and decision-making in the face of unexpected market fluctuations. Real data application is based on longitudinal U.S. data on real estate transactions.