Structuring Global Property Portfolios: A Cointegration Approach

Abstract

We create globally diversified real estate portfolios using cointegration methods over 1992-2009. Cointegration is robust to intertemporal correlation instability, identifies markets that share common factors and long-term trends, and identifies leading markets that do not respond to deviations from equilibrium within each cointegrated region. Our cointegration inspired model portfolios outperform the mean-variance optimized portfolio by 575-725 basis points annually. Differences in performance remain significant over separate time periods, and after controlling for various risk factors. Collectively, the results help clarify property portfolio selection and allocation policy vital for institutional investors and global real estate portfolio managers.

Keywords: performance, conitegration, allocation policy