Abstract
EXECUTIVE SUMMARY
This paper – the first of four working papers re-examining the role of real estate in mixed-asset portfolios for the
Investment Property Forum – reviews the literature on the relationship between real estate returns and those of
other asset classes.
The use of standard portfolio optimiser techniques using equity, bond and real estate indices tends to point
towards a large allocation to real estate – far greater than the proportion held by most professional investors in
their mixed-asset portfolios.
This result is driven by property’s reported low standard deviation and low correlations with the other financial
asset classes. However the allocation results hold even where property returns are desmoothed and illiquidity
premia are added.
Portfolio theory requires that asset returns follow a particular distributional form that means that risk can be fully
characterised using just individual asset standard deviation and the covariance between assets.
Research suggests that real estate does not meet the requirements of portfolio theory and the capital asset pricing
model: skewness and excess kurtosis lead real estate to fail tests for normality.
There is some evidence that return distributions are time-varying and that real estate is characterised by
asymmetric and/or non-linear behaviour, with differences observed between bull and bear markets. This suggests
that more complete risk management strategies are required.
In particular, the nature of property returns cast doubt on the use of the standard deviation as the most effective
measure of risk for property returns. Researchers have advocated the use of mean absolute deviation, downside
risk and lower partial moments, drawdown and other ways of capturing the risk of the asset class.
Much real estate performance data relies on valuations rather than transaction-based evidence. This is problematic
since it is believed that valuers anchor on past valuations and thus report smoothed, moving average, returns.
Methods that seek to desmooth valuation-based returns rely on a set of assumptions and there is no consensus as
to the most appropriate method.
Analysis of REIT returns suggests that the relationship between property and other asset classes is more complex
than portrayed in standard mean–variance approaches. The sensitivity of REITs to the overall stock market and to
bonds appears to be time varying or cyclical in nature, while the correlation between real estate and equities varies
over time. There is some evidence that correlations increase when markets are more volatile, with implications for
the benefits of diversification.
Studies employing a copula approach have identified a higher than expected probability of pairs of negative
returns, particularly between real estate and equity indices, which suggests the presence of lower-tail dependence.
These findings suggest that the diversification benefits of including real estate in a portfolio vary over time – and
may diminish in difficult market conditions.
Most of the results found relate to relatively high frequency data – daily, weekly or monthly. As short-run returns
are aggregated, the time varying effects become less evident. This might suggest that the impact of complex return
relationships is more likely to be felt in public real estate markets than in private property
This paper – the first of four working papers re-examining the role of real estate in mixed-asset portfolios for the
Investment Property Forum – reviews the literature on the relationship between real estate returns and those of
other asset classes.
The use of standard portfolio optimiser techniques using equity, bond and real estate indices tends to point
towards a large allocation to real estate – far greater than the proportion held by most professional investors in
their mixed-asset portfolios.
This result is driven by property’s reported low standard deviation and low correlations with the other financial
asset classes. However the allocation results hold even where property returns are desmoothed and illiquidity
premia are added.
Portfolio theory requires that asset returns follow a particular distributional form that means that risk can be fully
characterised using just individual asset standard deviation and the covariance between assets.
Research suggests that real estate does not meet the requirements of portfolio theory and the capital asset pricing
model: skewness and excess kurtosis lead real estate to fail tests for normality.
There is some evidence that return distributions are time-varying and that real estate is characterised by
asymmetric and/or non-linear behaviour, with differences observed between bull and bear markets. This suggests
that more complete risk management strategies are required.
In particular, the nature of property returns cast doubt on the use of the standard deviation as the most effective
measure of risk for property returns. Researchers have advocated the use of mean absolute deviation, downside
risk and lower partial moments, drawdown and other ways of capturing the risk of the asset class.
Much real estate performance data relies on valuations rather than transaction-based evidence. This is problematic
since it is believed that valuers anchor on past valuations and thus report smoothed, moving average, returns.
Methods that seek to desmooth valuation-based returns rely on a set of assumptions and there is no consensus as
to the most appropriate method.
Analysis of REIT returns suggests that the relationship between property and other asset classes is more complex
than portrayed in standard mean–variance approaches. The sensitivity of REITs to the overall stock market and to
bonds appears to be time varying or cyclical in nature, while the correlation between real estate and equities varies
over time. There is some evidence that correlations increase when markets are more volatile, with implications for
the benefits of diversification.
Studies employing a copula approach have identified a higher than expected probability of pairs of negative
returns, particularly between real estate and equity indices, which suggests the presence of lower-tail dependence.
These findings suggest that the diversification benefits of including real estate in a portfolio vary over time – and
may diminish in difficult market conditions.
Most of the results found relate to relatively high frequency data – daily, weekly or monthly. As short-run returns
are aggregated, the time varying effects become less evident. This might suggest that the impact of complex return
relationships is more likely to be felt in public real estate markets than in private property
| Original language | English |
|---|---|
| Publisher | Investment Property Forum |
| Publication status | Published - 2012 |