Can we use cap rates to better allocate investments in commercial real estate in a dynamic portfolio?

Abstract

This thesis has a two-fold objective, namely to explore the role of cap rates in predicting the returns to commercial real estate, and to identify how cap rates can be used to improve the allocation of real estate in a dynamic investment portfolio. Seeking an answer to the first question, we run predictive regressions using data for real estate "All Properties" and for all four major property types, examining the predictability power of cap rates for a forecasting horizon from one to four quarters in the future. Moreover, we examine whether or not stock dividend-price ratio can predict real estate returns, and examine the predictability of stock returns by cap rates and dividend-price ratio. The analysis confirms that both cap rates and the dividend-price ratio can predict real estate "All Properties" returns for up to one year in the future. Concerning the analysis per property type, the results vary from property type to property type, and for different forecast horizons. Moreover, the analysis shows that stock returns can be predicted by the dividend-price ratio at all forecast horizons, whereas the cap rates seem to have somewhat limited predictive power regarding the stock returns. We approach the second question by following the dynamic portfolio allocation methodology proposed by Brandt and Santa-Clara (2006). We expand the existing set of "basis" assets comprised of stocks and real estate to include "conditional" portfolios, and then compute the portfolio weights of this expanded set of assets by applying the Markowitz solution to the optimization problem. We apply this methodology to three different portfolio rebalancing horizons. Moreover, we work with three cases for each portfolio, i.e. with the unconditional case, with the case where the dividend-price ratio is the only conditioning variable, and with the case where the cap rate is the second conditioning variable. In almost all instances the results confirm that, by adding the cap rate as an additional state variable, the performance of the portfolios increases significantly. The same conclusion stands when we impose a "no shorting" restriction to real estate, although now the role of cap rates seems somewhat less significant.