Enhanced portfolio optimisation model for real estate investment in HK

Abstract

This paper investigates the role of direct real estate investment and securitised properties in a multi‐asset portfolio with financial assets available in Hong Kong, in a variety of time horizons. Grounded on the mean‐variance framework within the Modern Portfolio Theory, the study extends it by deploying the Exponentially Weighted Moving Average (EWMA) technique to estimate the variance and covariance, which reduces estimation errors, owed to the lack of ‘dynamic update’ capabilities in the standard model. Additionally, a constraint on the allocation to direct real estate investment is imposed in the portfolio optimisation problem to examine how percentage changes in the allocation to real properties affect the return and risk of the optimal portfolio. The experimental results show that the private domestic plays a more important role than property stocks in an optimal multi‐asset portfolio, in all time horizons, with allocation ranging from 23–27%. Also, for an optimal portfolio with shorter time horizons, lesser‐value direct real estate (Class A/B) tends to be included, while luxury property investment is the main asset for portfolios with longer time horizons. This proffers some implications for fund managers in Hong Kong in the management of portfolio investment when real estate is involved, subject to various levels of returns, risks and longevity.

Keywords:

• real estate investment
• private domestics
• exponentially weighted moving average (EWMA) technique
• portfolio management

Acknowledgements

This study was funded by the Hong Kong Polytechnic University’s Internal Grant (Project. #: ZZ1Z & G‐YH27). In addition, the authors would like to thank Mr Yu Ka‐hung for his assistance.

Notes

1. The component index (the index for a property class or grade) has been derived from analysis of all transactions effective in a given period. The composite index for a certain type of premises is compiled by calculating a weighted average of the component indices. For domestic premises the weights for both rental and price indices are based on the number of transactions effected in the current and previous 11 months.