31
To Hold or Not to Hold? An Analysis of Holding Periods in Five European Property Markets Jan Reinert [email protected] July 2013 Portfolio Analyst, IPD Germany PhD Candidate, University of Regensburg

To Hold or Not to Hold?

Embed Size (px)

DESCRIPTION

To Hold or Not to Hold?. An Analysis of Holding Periods in Five European Property Markets. Jan Reinert [email protected] July 2013 Portfolio Analyst, IPD Germany PhD Candidate, University of Regensburg. Agenda. Literature Data Analysis Average Holding Periods - PowerPoint PPT Presentation

Citation preview

Page 1: To Hold or Not to Hold?

To Hold or Not to Hold?

An Analysis of Holding Periods in Five European Property Markets

Jan [email protected]

July 2013Portfolio Analyst, IPD GermanyPhD Candidate, University of Regensburg

Page 2: To Hold or Not to Hold?

Agenda

• Literature

• Data

• Analysis• Average Holding Periods• Determinants of Holding Periods• Holding Periods & Performance

• Conclusion

Page 3: To Hold or Not to Hold?

Literature

Page 4: To Hold or Not to Hold?

LiteratureRESEARCHERS MARKET TIME

PERIOD METHOD HOLDING PERIOD OTHER FINDINGS

Collett, Lizieri& Ward(2003)

London, UK

1981-1998

Cox proportional

hazard model

7-12 years (median for

standard shops)

Decreasing holding periods over time,holding periods differ by size & sector,

negative relationship between return volatility & holding period

Gardner & Matysiak

(2005)

London, UK

1983-2003

Based on sold

properties

4.6-7.0 years (median over

time)

Decreasing holding periods over time,holding periods differ by location & investor type,

25% of properties are resold after 3 years,declining pattern of return over holding period

Brown & Geurts(2005)

San Diego, United States

1970-1990

Based on sold

properties, OLS

regression

4.5 years(average for apartment buildings)

Property characteristics (besides size) do not affect holding periods

Fisher & Young (2000)

United States

1980-1998

Time until 50% of

sample has been sold

8.6-13.7 years (median

depending on sector)

Decreasing holding periods over time,holding periods differ by sector,

returns converge to market average as tenure lengthens

Cheng, Lin& Liu

(2010)

United States

1978-2008

Model for ex ante optimal holding period

4.3-5.3 years (expected optimal

holding period)

higher transaction costs lead to longer holding periods while price volatility decreases it

Page 5: To Hold or Not to Hold?

Data

Page 6: To Hold or Not to Hold?

Data Set

...supplied by IPD:

TransactionInformation

(purchase & sale dates.price. transaction costs)

PropertyCharacteristics

(location. sector. age. size.market rent)

PerformanceMeasures

(total return. income return,capital value growth)

MarketPerformance

(National IPD Indices)

InvestorInformation(Investor Type)

Page 7: To Hold or Not to Hold?

Data Set 51.110 Observations... ...from 5 European Countries

Page 8: To Hold or Not to Hold?

AnalysisAverage Holding Periods

Page 9: To Hold or Not to Hold?

Average Holding Periods (all years)

Bottom 10%

Top 10%Upper Q.

Lower Q.

AverageMedian

Page 10: To Hold or Not to Hold?

Distribution of Observations (by year of purchase)

Page 11: To Hold or Not to Hold?

Problem:sample selection bias due to

different time periods of data sets

reducing datasets to the same time period:1995-2012

Page 12: To Hold or Not to Hold?

Average Holding Periods (1995-2012)

Bottom 10%

Top 10%Upper Q.

Lower Q.

AverageMedian

Page 13: To Hold or Not to Hold?

Problem:Holding periods are only observed for sold properties, held properties are ignored

”Survival Rates”

Page 14: To Hold or Not to Hold?

Median

50% of properties in Sweden were sold after 6 ½ years while 50% of properties in

the UK were sold after 9 years.

10% 20% 30% 40% 50%

UK 1.8 years 3.0 years 4.3 years 6.1 years 9.1 years

Germany 4.2 years 8.3 years

France 3.0 years 4.9 years 7.8 years

Netherlands 2.5 years 4.7 years 6.3 years 9.4 years

Sweden 1.3 years 1.9 years 3.1 years 4.5 years 6.5 years

Survival Rates (1995-2012)

After what time have XX% of properties been resold?

Page 15: To Hold or Not to Hold?

Survival Rates (1995-2012)

Page 16: To Hold or Not to Hold?

Holding Periods over Time (by year of purchase)

Page 17: To Hold or Not to Hold?

Holding Periods over Time (by year of sale)

Page 18: To Hold or Not to Hold?

AnalysisDeterminants of Holding Period length

Page 19: To Hold or Not to Hold?

Average Transaction Costs (1995-2012)

Page 20: To Hold or Not to Hold?

Risk/Return Profiles (1998-2012)

1-for-1-trade-off

Page 21: To Hold or Not to Hold?

HPMONTHSINSURANCE 3.4**

PENSION 11.9***OPENFUND 4.8***

CLOSEDFUND 12.7***LISTEDCOMP 3.4*

UNLISTEDCOMP 1.3PRICEPURCH -0.0

PRICESALE 0.0***TCPURCH 0.1

TCSALE -0.0TR -2.9***RR 3.3***

STDEVRR -0.2***NETHERLANDS 18.8***

SWEDEN -46.5***FRANCE 23.6***

GERMANY -6.2**CONSTANT 84.3***

R2 0.22

OBS 9,723

Regression HPMONTHS

Y1995 22.7***Y1996 23.3***Y1997 14.6***Y1998 10.6***Y1999 5.0***Y2001 -4.8***Y2002 -1.6Y2003 -5.8***Y2004 -11.7***Y2005 -22***Y2006 -36.6***Y2007 -47.5***Y2008 -50.1***Y2009 -42.9***Y2010 -42.8***Y2011 -55***

METRO 0.6OFFICE -5.8**RETAIL -3.7

INDUSTRIAL 0.3RESIDENTIAL 8.2***

Page 22: To Hold or Not to Hold?

Regression UK NETHERLANDS SWEDEN FRANCE GERMANY

Y1995 23.1*** 50*** Y1996 24.7*** 43.6*** Y1997 18.1*** 37.9*** 43.2*** -1.1***Y1998 12.3*** 34.7*** 29.3*** 8.4Y1999 8.5*** 14.7*** -3.2 -33.5*** 14.1Y2001 -6.3*** -17.3*** -49.3*** -7.7 -4.6*Y2002 -5.3*** -26.2*** -99.1*** -10.6 -12.4Y2003 -4** -23.5*** -28*** -36.5*Y2004 -9.4*** -29.7*** 29.6 -47.7*** -27***Y2005 -26.1*** -22.1*** -8.5 -52.3*** -43.7***Y2006 -50.2*** -34.2*** -20.2* -57*** -40.9***Y2007 -75.5*** -73*** -47.2*** -79.8*** -48.1***Y2008 -82.8*** -158.1*** -69.2*** -91.1*** -60.7***Y2009 -46.9*** -67.7***Y2010 -35.7*** -52.9*** -73.7***Y2011 -60.2***

METRO 0.2 3.7 1 0.7 4.7***OFFICE -9.7*** -29.7*** -25.7** -18.4** 11.1RETAIL -9.6*** -6.8 -10 -0.2 1.2*

INDUSTRIAL -1.8 -10.8 0.4 -24.8*** 5.2RESIDENTIAL 17*** -8.4 3.9 -21.4* -15.9INSURANCE 2.2 -18.3*** 2.8 14.2** -25.4*

PENSION 10.5*** 10.8* 8.5 -94.5***OPENFUND 1.2 -3.8 18.2* -16.2***

CLOSEDFUND 4.7 0.5 -1**LISTEDCOMP -1.8 11.4** 15.7*** -27

UNLISTEDCOMP 0.8 -10.5*** 13.5* -3.3 -0.6*PRICEPURCH 0*** 0 0* 0 0

PRICESALE 0*** 0*** 0*** 0 0***TCPURCH 0 0 3.9*** 1.2** 0.4***

TCSALE -0.2 -0.6 -1.8** 2.9*** 0.3TR -4.2*** -13.5*** 7.4*** -3.9*** 10.8RR 4.8*** 15.9*** -7.4*** 4.1*** -11.1***

STDEVRR -0.2*** -0.4** -0.1 0 -0.3***CONSTANT 103.5*** 233.6*** 134.7*** 131.8*** 56.1

R2 0.24 0.58 0.49 0.51 0.67

OBS 8257 591 393 330 152

Page 23: To Hold or Not to Hold?

Heckman Correction…in progress…

Problem:Censored data – Holding periods are only observed for sold properties, held properties are ignored

Page 24: To Hold or Not to Hold?

AnalysisHolding Periods & Performance

Page 25: To Hold or Not to Hold?

Holding Period & Excess Return (1995-2012)

Page 26: To Hold or Not to Hold?

Holding Period & Excess Return (all years)

Page 27: To Hold or Not to Hold?

Conclusion

Page 28: To Hold or Not to Hold?

Conclusion I• Holding periods seem to differ by country: between 1995-2012 the

Netherlands had the longest while Sweden had the shortest simple average holding period

• Survival rates show that after 9 years 56% of properties in Sweden had been sold again while only 21% of properties in Germany had been resold

• 50% of properties in the UK and Sweden had been resold after 9 and 6 ½ years respectively

• Over the period of analysis less than 50% of properties in the Netherlands, France and Germany had been resold

• Holding periods in the UK seem to be declining since 1980 but the same cannot be said for the Netherlands and Sweden

Page 29: To Hold or Not to Hold?

Conclusion II• Transaction costs were lowest in Sweden (which also had the shortest

average holding period)

• France had the highest average transaction costs (especially for sales)

• Between 1998-2012 the UK market displayed the highest volatility while Germany was the least volatile (Sweden displayed the 2nd highest volatility)

• A regression over all countries showed that transaction costs did not influence holding periods (differening results in individual countries)

• Relative return had a positive impact on holding periods while return volatility had a negative impact

• In line with other studies, a pattern of declining excess return over holding period length was identified

Page 30: To Hold or Not to Hold?

Conclusion IIIHolding period analysis is constrained by the problem of

CENSORED DATAHolding periods of properties still held by investors are

unobservable.

This can explain:

a) differing average holding periods across countries

b) declining holding periods over time

c) pattern of declining excess return over holding period

Page 31: To Hold or Not to Hold?

Thank you for your attention.

Jan [email protected]

Portfolio Analyst, IPD GermanyPhD Candidate, University of Regensburg