Efficiency in the City of London office market : A supply perspective

Real Estate & Planning: www.henley.ac.uk/repwww.henley.ac.uk/rep

Steven Devaney (University of Reading), Oliver Holtemöller (Halle Institut for Macroeconomics) and Rainer Schulz (University of Aberdeen)

Efficiency in the City of London office market: A supply perspective

Real Estate & Planning: www.henley.ac.uk/rep

• Why does it matter?– Land use allocation within property market– Resources allocated to property in the

economy– Investment flows of financial institutions

• Our objectives:– Test the informational efficiency of prices

(yields) in the City of London office market– Explore whether mispricing affects office

development decisions

Informational efficiency


• Analyses of prices or yields might– Test whether they react as expected to

changes in fundamental drivers– Estimate ‘rational’ prices or yields and examine

how these differ, e.g. Hendershott (1996, 2000)

• Findings are often against efficiency, but common issues are– Quality of data and appraisal basis of data– Role of expectations

Previous literature


Sivitanides et al. (2001)

• Regress NCREIF cap rates onto a priori determinants

• Movements not rational given mean reversion in rents

Chen et al. (2004)

• Regress spread over bond rate onto a priori determinants

• Movements also not rational, though authors try and justify

Hendershott& MacGregor (2005a)

• Extensive cleaning of NCREIF data before modelling

• Still found irrational relationships with income growth proxies

Previous studies – US


McGough & Tsolacos (2001)

• IPD property yields cointegrate with gilt and dividend yields, but not rent growth, whilst ECM part doesn’t work

Hendershott& MacGregor (2005b)

• Property yields cointegrate with proxies for cash flow growth and equity market variables

• Results suggest that UK cap rates have been rational

Clayton et al. (2009)

• Use survey data on risk premiums and expected rent, plus sentiment indicators

• Argued that fundamentals are the main driver of US cap rates over time

Previous studies – UK & US


• We construct ‘rational’ multipliers (1 / yield) and compare these with actual multipliers

• Based on well known approach of Campbell & Shiller (1988) for equity market

• Start with expression for present value:

Our approach

1j1t

ktj1k

jtt Ω

)H(1ΠD

EP


• Expressed in terms of multipliers:

• We model what the income multiplier rationally should be given information on key inputs

• But expectations and required return rates are not observed directly

Our approach

1j1t

ktj1k

itj1i

t Ω)H(1Π)G(1ΠEM


• Use VAR to forecast inputs given information on their past values and those of related variables

• Use four different assumptions on how required returns are set:a. Constant in nominal termsb. Constant in real termsc. Risk premium is constantd. Linked to returns on other risky assets

Our approach


• We examine 1952-2012– Office market data: rents and yields (Devaney,

2010; Scott, 1996; CBRE)– Financial data: equity returns and yields, gilt

returns and yields (Barclays Capital, 2013)– Economic data: GDP growth and inflation (ONS)– Development data: stock and completions

(Smyth, 1985; Barras, 1979; City of London), construction costs (BCIS, ONS)

Dataset


Sources: Office initial yields – Scott (1996), CBRE. Gilt yields and dividend yields – Barclays Capital (2013)


Case A Case B Case C Case DLag length 1 1 1 1St. dev. ratio (m/m*)

1.03 0.53 0.41 0.22

Multiplier correlation

-0.07 -0.20 -0.29 -0.25

LR test statistic 4.344 4.928 5.484 27.133p-value 0.36 0.29 0.24 0.00

Results

• High p-values mean that efficiency cannot be rejected

• However, graphs reveal sustained differences between simulated and actual multipliers

Simulated vs. actual multiplier


1952195419561958196019621964196619681970197219741976197819801982198419861988199019921994199619982000200220042006200820102012

-0.30

-0.20

-0.10

0.00

0.10

0.20

0.30

0.40

0.50Case A Case B Case C

Obse

rved

– S

imul

ated

log

mul

tiplie

r

Required return assumptions: A = constant in nominal terms, B = constant in real terms, C = constant risk premium, D = varies with equity returns


• Second (structural) VAR to explore this aspect

• Inputs: completions, costs, simulated multiplier and estimated mispricing term

• Impulse response functions indicate if shocks in one variable (e.g. mispricing term) subsequently affect others (e.g. completions)

• Potential interpretations of responses are strategic behaviour or shared (wrong) perceptions

Developer response


SVAR output


• Initial finding: informational efficiency cannot be rejected, but sensitive to model and lags

• Work is in progress to check the stability and the sensitivity of models and results

• SVAR results are suggestive of developer response to instances of mispricing

• Related work is in progress with regard to pricing of real estate equities and manager responses

Conclusions and issues


Sources: Rent – Devaney (2010), CBRE.Stock – our estimates, City of London local authority.


1952

1955

1958

1961

1964

1967

1970

1973

1976

1979

1982

1985

1988

1991

1994

1997

2000

2003

2006

2009

2012

0.0

0.2

0.4

0.6

0.8

1.0

1.2

1.4Case A Case C Actual

Log

Multi

plie

rs (n

orm

alise

d)

Required return assumptions: A = constant in nominal terms, B = constant in real terms, C = constant risk premium, D = varies with equity returns


The authors are grateful for permission from CBRE to use their unpublished historical rent and yield series in the analysis and to Barclays Capital for permission to use data from the Equity Gilt Study 2013.

An earlier version of the paper can be obtained fromhttp://www.iwh-halle.de/e/publik/disc/15-12.pdf

Contact : Dr Steven Devaney ([email protected])

18

http://www.iwh-halle.de/e/publik/disc/15-12.pdf

http://www.iwh-halle.de/e/publik/disc/15-12.pdf

mailto:[email protected]

Efficiency in the City of London office market : A supply perspective

Documents

onsdatasetreal estate

approachreal estate

ukrepreal estate planning

dividend yields

office market data

economic data

cbrefinancial data

use survey data