Exploring a new technique to determine the optimal real estate portfolio allocation

Abstract

Modern Portfolio Theory has been developed over the last fifty years, and there are several studies linking Modern Portfolio Theory with the allocation of real estate property in multi-asset portfolios. However, in reality, most real estate fund managers don't use MPT as a guideline when they are structuring a portfolio and deploying allocation strategy for a real estate fund. The main reason for this gap between theory and reality is that the traditional mean-variance approach of MPT requires accurate data of variances, covariance and expected return over the long term; and those data are quite difficult to collect on an ad hoc base. This Thesis applies a new technique to examine property asset allocation strategies and improve the performance of a real estate investment sample portfolio in the US. We straight-model the portfolio weight in each property type of asset as a function of the asset's characteristics: either physical attributes such as property size, vacancy rate, property type, location etc.; or financial attributes such as Cap Rate. The coefficients of this function are found by optimizing the investor's average utility of the portfolio's return over a certain period of years. The aim of this approach is to find a simple and easily modified methodology for real estate portfolio managers when they are deciding on acquisitions and making portfolio policies. In general, this Thesis aims to apply the new technique to help practitioners and other researchers improve the practical implementation of optimal portfolio policies.