Exploring the determinants of liquidity with big data – market heterogeneity in German markets
| DOI | https://doi.org/10.1108/JPIF-01-2017-0006 |
| Published date | 05 February 2018 |
| Date | 05 February 2018 |
| Pages | 3-18 |
| Author | Marcelo Cajias,Philipp Freudenreich |
| Subject Matter | Property management & built environment,Real estate & property,Property valuation & finance |
Exploring the determinants of
liquidity with big data –market
heterogeneity in German markets
Marcelo Cajias
Department of Research, PATRIZIA Immobilien AG, Augsburg, Germany, and
Philipp Freudenreich
IRE/BS International Real Estate Business School, University of Regensburg,
Regensburg, Germany
Abstract
Purpose –The purpose of this paper is to examine the market liquidity (time-on-market (TOM)) and its
determinants, for rental dwellings in the largest seven German cities, with big data.
Design/methodology/approach –The determinants of TOM are estimated with the Cox proportional
hazards model. Hedonic characteristics, as well as socioeconomic and spatial variables, are combined with
different fixed effects and controls for non-linearity, so as to maximise the explanatory power of the model.
Findings –Higher asking rent and larger living space decrease the liquidity in all seven markets, while the
age of a dwelling, the number of rooms and proximity to the city centre accelerate the letting process. For the
other hedonic characteristics heterogeneous implications emerge.
Practical implications –The findings are of interest for institutional and private landlords, as well as
governmental organisations in charge of housing and urban development.
Originality/value –This is the first paper to deal with the liquidity of rental dwellings in the seven most
populated cities of Europe’s second largest rental market, by applying the Cox proportional hazards model
with spatial gravity variables. Furthermore, the German rental market is of particular interest, as
approximately 60 per cent of all rental dwellings are owned by private landlords and the German market is
organised polycentrically.
Keywords Non-linearity, Big data, Cox proportional hazards model, Housing real estate,
Liquidity/time-on-market, Spatial effects
Paper type Research paper
1. Introduction
Financial assets such as stocks and bonds are traded in tremendous volumes, turning over
billions of dollarswithin seconds and with no spatial constraints.By contrast, the transaction
process of directreal estate is more complex, often consumingseveral months. When it comes
to residential real estate, a match may be even more difficult to achieve, as this is strongly
determined by theindividual preferences of homebuyersand the expectations of homesellers.
A general understanding of liquidity in direct real estate is therefore essential for market
players, whetherprivate, institutional orgovernmental, not only in order to deriveinvestment
strategies,but also to assess market fundamentals and cyclicalmovements, as well as political
implications.Moreover, the instrumentsneeded to efficiently capturethe factors both boosting
and constraining liquidity, are crucial and far from trivial, as liquidity in terms of “time”
requires advanced econometric modelling. To capture the uncertainty of finding a match, as
well as the time a propertyis advertised on the market, liquidity in the residential real estate
literature is widelyproxied by time-on-market (TOM). In this context, this paper exploresthe
liquidity of direct real estate, focussing on the seven largest German rental housing markets
by means of advancedsemi-parametric survivaltechniques. The aim of the studyis to explore Journal of Property Investment &
Finance
Vol. 36 No. 1, 2018
pp. 3-18
© Emerald PublishingLimited
1463-578X
DOI 10.1108/JPIF-01-2017-0006
Received 18 January 2017
Revised 22 March 2017
Accepted 24 March 2017
The current issue and full text archive of this journal is available on Emerald Insight at:
www.emeraldinsight.com/1463-578X.htm
The authors especially thank PATRIZIA Immobilien AG for contributing the data set for this study.
All statements of opinion are those of the authors and do not necessarily reflect the opinions of
PATRIZIA Immobilien AG or its associated companies.
3
Market
heterogeneity
in German
markets
liquidity concepts and examine the factors that determine liquidity, such as linear, binary,
spatial as well as possible non-linear effectswith big data, in order to derive both similarities
and divergences between the cities. The paper may serve as a guide for market players and
policy makers conducting liquidity analysis on and understanding future developments in
rental housing markets in Germany. Especially for first-time buyers, an overview on the
largest seven realestate markets and an indication of thefactors affecting the letting process
is of considerable importance, as during the marketing time, redemption and interest have to
be borne by other sources of income.
The following brief literature review only coversthe articles directly relevantfor this study,
and thus only a small fraction of the literature on TOM. Since their establishment within the
real estate literature, survival models have been adopted by various researchers to estimate the
determinants of TOM. Kluger and Miller (1990) introduced the semi-parametric Cox
proportional hazards model based on Cox (1972) to real estate studies, which allows a
particularly flexible application, without any a priori assumptions regarding the distribution of
the baseline hazard, in contrast to the widely used Weibull model. Studies using this approach
include Krainer (1999), Smith (2010), Hoeberichts et al. (2013), Cirman et al. (2015), among others.
In searching for an instrument capturing “user taste”for dwellings and its effect on
liquidity, Haurin (1988) developed an atypicality index and shows that for more atypical
dwellings, the distribution of offers is prone to wider variation. A dwelling is more atypical
when its hedonic properties deviate substantially from the mean hedonic market
characteristics, e.g. a dwelling with 150 m
2
, one room, located on the tenth floor without an
elevator. Nowadays, atypicality is a widely recognised factor in hedonic survival
regressions, as seen in Krainer (1999), Anglin et al. (2003), Bourassa et al. (2009), Haurin et al.
(2010, 2013) and Hoeberichts et al. (2013), among others.
The signalling effect of setting the initial list price is also a widely researched area.
Glower et al. (1998) for example, began to investigate the impact of the percentage difference
in the observed list price from the expected list price. Anglin et al. (2003) extended this
approach and introduced a new explanatory variable in the context of liquidity, called the
degree of overpricing (DOP). They defined the variable as the percentage deviation of an
individual property’s list price from the empirically estimated market list price. While they
found that abnormal list prices, i.e. overpricing, increase the marketing time of houses,
further applications can also be found in Hoeberichts et al. (2013) and Cirman et al. (2015).
Over the last years, more and more emphasis has been placed on spatial effects when
modelling theprice and TOM of residential properties.Many articles have tested the theoryof
market segmentation in residentialreal estate markets, concludingthat the inclusion of spatial
variables improves the explanatory power of real estate pricing models, e.g. Goodman and
Thibodeau (2007), Turnbull and Dombrow (2006), Pavlov (2000), Fik et al. (2003), Bourassa
et al. (2010) and Cirman et al. (2015) among others. Smith (2010) was the first to specify a
Cox-model containing school districts and Cartesian coordinates. He found that, while the
school district dummies and the coordinates are by themselves statistically significant and
demonstrate a large impact on the liquidity, the combination of both, yields the largest
explanatory power.
To the best ofour knowledge, the first studyto estimate TOM in residentialrental markets
was conducted by Allen et al.(2009). The authors focus on the Dallas-Fort Worth area with a
sample of over 20,000 listings and more than 11,000 corresponding letting contracts. Using a
Weibull hazard model, the authors conclude that after resetting asking rent initially
overpriced by 15 per cent, the landlords face 9.5 days longer TOM on average. Due to the
initial overpricing and thus longer TOM, these landlords also have to accept a contract rent
which is on average 5.2 per cent below the hedonically estimated level. Cajias et al. (2016),
in contrast, used a similar approach to estimatethe effect of energy consumption on TOM for
the German rental market. Using a Cox model, the authors calculated the odds of a dwelling
4
JPIF
36,1
To continue reading
Request your trial