Measuring sentiment in real estate – a comparison study

Pages248-258
Published date03 April 2018
DOIhttps://doi.org/10.1108/JPIF-05-2017-0034
Date03 April 2018
AuthorSteffen Heinig,Anupam Nanda
Subject MatterProperty management & built environment,Real estate & property,Property valuation & finance
Measuring sentiment in real
estate a comparison study
Steffen Heinig and Anupam Nanda
Henley Business School, University of Reading, Reading, UK
Abstract
Purpose In mainstream economics and finance literature, market sentiment is considered irrational.
This leads to significant challenges in capturing the effect of sentiment on economic relationships. Real estate
is even more complex due to the fact that the sector exhibits several market inefficiencies. The purpose of this
paper is to explore the literature and present a simple test for the potential of using three different sentiment
indicators to improve a basic cap rate model. The authors establish the case using commercial real estate
(CRE) data for London West End.
Design/methodology/approach The three indicators differ in their underlying source and method.
The authors used orthogonalisation and principal component analysis for a macroeconomic sentiment indicator.
Furthermore, online search volume data have been used to mirror the market sentiment for the London West End
market. Finally, textual analysis based on word lists has been applied to corpus of market reports.
Findings The results indicate considerable improvement in the authorsability to capture the effect of
sentiment. Furthermore, the consideration of a human factor leads to improvement in the basic yield model.
Practical implications The methods suggest that sentiment extracted from more forward-looking
sources, such as online searches, could be a significant information gain for investors, lenders or other market
participants. The additional information could be used to adjust their behaviour within the market.
Originality/value To the authorsknowledge, this is the first study that applies textual analysis to market
reports for the CRE market in the UK.
Keywords Sentiment, Linguistics, Textual analysis, Natural language, Property lending, Yield modelling
Paper type Research paper
1. Introduction
Understanding the role of investor sentiment is not new to economics and finance literature.
While the role of sentiment is well understood but its measurement and impact under
complex, informationally asymmetric economic environments are not straightforward, as it
can differ for different types of agents at various times across different asset classes.
Any causal analysis will, therefore, suffer from substantial endogeneity issues leading to
biased estimates and wrong statistical inferences. This is especially problematic in
commercial real estate (CRE) due to severe information asymmetry. The nature of the global
real estate investment, especially due to the lessons learnt during the Global Financial Crisis,
requires a deeper understanding of the risk-taking behaviour and the associated
determinants. Much of the volatility can be captured through examining cap rate or yield, in
which sentiment can play a significant role. Sentiment can lead to quick, major changes in
risk premia and subsequent deflation in property values.
The study of sentiment is profoundly under-researched in real estate pricing models and
the existing real estate investment literature. In this study, we demonstrate how
the capturing sentiment in yield model can significantly improve explanatory power.
We use the London West End office real estate market as our testing ground. We present
three different ways of extracting sentiment in this paper and apply the sentiment indicators
to a standard yield model. The first method uses indirect sentiment proxies, which mirror
the market development through macroeconomic indicators. The second method utilises
online search volume data, as a tool to approximate the beliefs of market participants.
The last procedure presents a new and convenient way to mirror the market sentiment.
We apply computational linguistic methods to extract the sentiment from various market
reports. Our results suggest that the models benefit from the additional information which
Journal of Property Investment &
Finance
Vol. 36 No. 3, 2018
pp. 248-258
© Emerald PublishingLimited
1463-578X
DOI 10.1108/JPIF-05-2017-0034
Received 2 May 2017
Revised 30 September 2017
Accepted 1 October 2017
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/1463-578X.htm
248
JPIF
36,3

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT