Private Commercial Real Estate Returns and the Valuation Process

EXECUTIVE SUMMARY
This paper is the second of four working papers re-examining the role of real estate in mixed asset portfolios for
the Investment Property Forum. In it the behaviour of valuation-based measures of the commercial property market
are examined. In particular, the extent to which the behaviour of real estate returns and the valuation of property
vary in different economic states is investigated.
The thinly traded commercial real estate market relies on periodic valuations to construct market performance
measures. Valuers draw on historic transaction data and prior valuations as a basis for their appraisals. As a result,
valuation-based return series will tend to lag the underlying market and dampen reported volatility in the market
– valuation smoothing. Smoothing effects are much more pronounced in high-frequency return series such as
monthly indices, where valuers tend to undertake desk-based valuations and update prior valuations. Over longer
time periods, the effects diminish.
Uncritical use of valuation-based series in portfolio allocation models could lead to sub-optimal weightings to
property. Techniques exist to desmooth property indices – attempting to estimate the true underlying pattern of
returns in the market. However, standard desmoothing techniques typically assume that smoothing effects are
constant over time and make strong assumptions about return processes.
Research in real estate and other asset markets suggest that both underlying market behaviour and valuer
behaviour may vary over the market cycle. This suggests that analysis based on the existence of distinct market
regimes – states of the world determined by particular economic conditions – may be helpful in understanding the
dynamics of property returns.
The research uses econometric approaches to test whether both the return process and the valuation smoothing
process vary over time, dependent on the state of the economy or wider market conditions. The technique used is
a Threshold Autoregression or TAR model. These models test whether the returns can be better characterised as
deriving from processes which differ according to economic state – for example, if the economy is in recession or
is booming.
The results of two TAR models are reported: a model where valuation smoothing is assumed to be constant across
time but the return process varies according to market conditions; and a second model, where both valuation
smoothing and underlying return processes are assumed to vary according to market conditions.
Commercial real estate performance is measured using the IPD Monthly Index. To match with some of the
aggregate macro-economic variables, this was aggregated to quarterly returns. Initially, data from 1986 to 2008
were analysed, to take in the onset of the property market correction. They were then reanalysed the data to mid-
2011, to provide robustness checks and to assess the continuing impact of global financial instability.
A number of factors thought to drive commercial property returns were tested. An output measure of GDP and
changes in service employment was used to test whether behaviour varied over the economic cycle. Returns from
the FT All Share Index were used to investigate linkage between equity returns and the property market. The
impact of interest rates was analysed, using three-month LIBOR as a reference value. ONS’s RPIX index was used
to test property–inflation linkages. Given the growing importance of foreign investment in real estate, the research
included an exchange rate variable, the US$ rate against sterling. Finally, as an internal indicator of the property
market, a yield or cap rate measure was included: after testing, IPD’s All Property initial yield figure was utilised