The investment risk of institutional-grade commercial real estate in Australia

Knowledge of the investment risk of investment-grade commercial real estate (‘ICRE’) is important because it determines the approaches which should be taken to portfolio management. However, relatively little is known about this risk. This research expands the body of knowledge of ICRE investment risk by producing conclusions about the information content of prices and the distribution of returns in the ICRE context. It is broken into three main parts. First, the ICRE returns-generating process is characterised to form a basis for deducing theoretical conclusions about the information content of prices and the stochastic attributes of returns. The rationale for this approach lies in capital markets literature, which demonstrates that the characteristics of the information structure of markets, the decision-making processes of investors and the market trading mechanism determine the main attributes of the process of price evolution (which is assumed to be the main driver of returns). The analysis concludes that ICRE prices are partially informed, and changes in prices are described by a ‘jump’ process. Second, analysis of a database of ‘large’ price changes supplied by the Property Council of Australia is undertaken to empirically test the jump process hypothesis. This analysis provides evidence that natural events associated with changes in the leasing structure of properties are a primary driver of relatively large, infrequent dislocations in valuation-based prices. With parts one and two as a backdrop, the third part of this research empirically tests a discrete mixture of normals (‘DMON’) model of investment risk. Capital markets research shows that a DMON model flows naturally from jump price processes. DMON models fitted to cross-sectional returns on individual properties supplied by the PCA are found to be superior to the normal and stable Paretian models previously proposed by other researchers. In aggregate these conclusions have serious implications for the management of ICRE portfolios, and suggest a need for additional research. Some implications include: (1) Mean-lower partial variance is superior to mean-variance optimisation. (2) Forecasting the distribution of ICRE returns forms a new tool for active management. (3) Passive portfolio management is inappropriate. (4) Comparables-based valuations may be unreliable for investment decisions.