It is well established that the volatility and correlation of return rates of many financial assets varies over time. Moreover, it is often suggested that the response of real estate return rates to risk factors might be different to that of other assets, meaning that real estate’s diversification benefit might vary over the economic cycle.

With funding provided by the RICS Education Trust, Liang Peng of the University of Colorado at Boulder, USA and Rainer Schulz of the University of Aberdeen, Scotland, used daily return rate data of real estate companies and other companies from 14 countries and regions between 1990 and 2007 to explore if time-variation exists, if it is asymmetric, and if it should be taken into account in portfolio risk management.

What did they find? It seems that the volatility and correlation of real estate and other stocks’ return rates are time-varying and that better performing portfolios could be created if investors were able to forecast the time-variation.

The study shows that recently developed time series models fitted with past information can be used to forecast the time-varying correlation. In an out-of-sample experiment they applied these time series models to restructure portfolios on a daily basis and found that the performance of these portfolios was better than portfolios formed by ignoring time-variation.

As Liang Peng and Rainer Schulz conclude, “Investors should be aware of the implications of time-varying correlation and that using a constant covariance matrix may misrepresent the true appeal of real estate investments.”