Frontiers of commercial real estate portfolio performance: Are sector-region-efficient diversification strategies a myth or reality?

Abstract

This paper departs from the traditional optimisation methods used to evaluate portfolio performance. Rather, the Stochastic Frontier Analysis approach is used to econometrically determine the benchmark real estate portfolio frontier and subsequently assess the gains from diversifying real estate portfolios along regional and sectoral dimensions in the UK. Portfolio specific inefficiency measures are obtained which indicate whether a portfolio is efficiently diversified and therefore places on the benchmark frontier and if not, the degree to which performance can be improved is quantified. Portfolio-specific efficiencies average at 85–91%, indicating scope to further improve performance. Further, diversification be it on a sectoral or regional dimension, contributes to significantly lower variability in portfolio efficiencies.

Keywords:

• Diversification
• portfolios
• performance
• real estate
• sector and regional effects

Acknowledgements

The authors thank Dr Peter Hayes, Managing Director Global Head of Investment Research PGIM Real Estate for generously providing the data-set used in this study.

Notes

1. A detailed analysis of the time-series features of the UK commercial property returns can be found at Coleman and Leone (2015).

2. An alternative approach, in the form of the Battese and Coelli (1995) SFA model, specifies the mean of inefficiency as a function of determinant variables. This requires a truncated normal distribution for the inefficiency term. MLEs for this specification failed to converge.

3. As noted by Kumbhakar and Lovell (2000), heteroscedasticity in the random error yields consistent estimates of the frontier parameters with the exception of the intercept. However, the resulting efficiency estimates are biased. Heteroscedasticity in the inefficiency component biases both the frontier parameter and the inefficiency estimates.

4. The sum of money spent on purchases of new properties, expenditure on development and other capital expenditure, or received through sales. Sales include whole or part sales and other capital receipts.

5. The sum of rent receivable plus other revenue receipts net of property-specific management costs, ground rents and other irrecoverable expenditure.

6. In addition to GDP and GDP growth rate, additionally, estimations were carried out using inflation (often hedged against using real estate) as a variance determinant. The estimations failed to converge.

7. The UK commercial property crashed in early 1990s. Between 1989 and 1993, UK commercial property prices fell by 27%.

8. See Grover and Grover (2013) for a comprehensive discussion about property cycles.