appraisal within open-end real estate funds: evidence on biased appraisals in fund crisis year 2006...
TRANSCRIPT
Appraisal within open-end real estate funds:Evidence on biased appraisals in fund crisis year 2006
Sebastian Glaesner
- Contact Author -
Doctoral CandidateEuropean Business School International University Soehnleinstr. 8/D
65201 Wiesbaden
2Sebastian GläsnerERES Conference 2009
The Role of GOEFs in the German Retail Market
GOEFs play with a fund volume in excess of 80 billion € major role in German retail market
Real estate held in trust by an investment company (KAG) Investments in commercial and residential properties as well as developments Constant issuance and redemption of units Unlimited number of unit owners First open-end real estate fund set up in Germany in 1959 by the Bayerische
Immobilien- und Wechselbank and the Bayerische Vereinsbank First legal regulation in 1969 The number of funds in Germany has increased from 8 (1972) to 43 (12/2008) Open-end real estate funds exist in Switzerland, Germany, Luxemburg and
France as well as in various other European countries
3Sebastian GläsnerERES Conference 2009
Low, but extremely stable nominal returns (1998-2007, by fund)
5.12%4.92% 4.79%
4.61% 4.57%
4.15%3.90% 3.89% 3.78%
3.34% 3.28%
1.96%
4.02%
0.0%
1.0%
2.0%
3.0%
4.0%
5.0%
6.0%S
EB
_Im
moI
nves
t
grun
dbes
itz_e
urop
a
CS
_EU
RO
RE
AL_
A_E
UR
HA
US
_IN
VE
ST
_eur
opa
Dek
a_Im
mob
ilien
Eur
opa
Uni
Imm
o_E
urop
a
Uni
Imm
o_D
euts
chla
nd
Wes
tInv
est_
1
HA
NS
Aim
mob
ilia
Dek
a_Im
mob
ilien
Fon
ds
DE
GI_
EU
RO
PA
EU
RO
_Im
moP
rofil
OF
IX-1
0
mean
sd
rf mean 1998-2007
SEB ImmoInvest achieved the highest returns in the ten year horizon and at the same time realised those returns with the lowest volatility of merely 0.61%. The majority of the remaining funds show an annualized return volatility slightly below one percent. As already mentioned in the introduction, grundbesitz europa has by far the highest volatility of 2.13% due to its high returns of 4.5% in February and 4.6% in March 2007
4Sebastian GläsnerERES Conference 2009
Explanations for stable or smoothed returns
product characteristics stabilizing returns stable cash flows from long lease duration diversification across Europe / worldwide investment in cash and bonds
product characteristics concealing return volatility twelve month valuation cycle lags market developments
potential influence other than market developments on return volatility
intentional valuation smoothing/ influence (dissertation paper 2) intentional timing of valuations to smooth monthly fund returns (dissertation
paper 3)
5Sebastian GläsnerERES Conference 2009
Strong yield compression in 2006
5.27
5.15
4.57
4.76
5.75
5.00
5.62
5.465.38
4.92
4.544.624.70
4.00
4.20
4.40
4.60
4.80
5.00
5.20
5.40
5.60
5.80
6.00
2004 2005 2006 2007
AtisReal BulwienGesa CBRE DEGI
The figure shows average (not capital weighted) prime yields of the top 5 German office locations Berlin, Düsseldorf, Frankfurt, Hamburg and Munich as reported by major real estate consultants. All four companies report the strongest yield compression in 2006.
6Sebastian GläsnerERES Conference 2009
Strong capital redemptions among many funds in crisis years 2005/2006
FV as % 2004-01-31 fund name
FV to 2004-01-31 2004-12-31 2005-12-31 2006-12-31 2007-12-31
AXA Immoselect 296,500 157% 418% 576% 1008% DEGI INTERNATIONAL 381,700 125% 248% 357% 482% Deka-ImmobilienEuropa 9,308,400 2% -10% -26% -27% Deka-ImmobilienFonds 7,047,600 -14% -32% -55% -63% Deka-ImmobilienGlobal 1,929,300 5% 2% -10% -13% grundbesitz-europa 8,332,600 -13% -32% -67% -72% grundbesitz-global 2,175,900 36% 30% 30% 65% HAUS-INVEST europa 11,917,700 -12% -28% -31% -18% KanAm grundinvest Fonds 1,100,000 77% 184% 159% 284% SEB ImmoInvest 4,420,100 13% 21% 31% 59% UBS (D) Euroinvest Immobilien 1,278,400 44% 48% 66% 78% UniImmo: Deutschland 9,353,400 -13% -33% -45% -46% UniImmo: Global 510,700* 125% 163% 236% WestInvest 1 3,280,400 -4% -15% -35% -43% WestInvest InterSelect 1,642,300 37% 55% 56% 55% *FV to 2005-01-31 The table shows fund volumes (FV, as net asset values in EUR) as beg. 2004. The subsequent columns show changes of fund volume relative to 2004.
7Sebastian GläsnerERES Conference 2009
German properties of open-end funds 2005-2007
2005 2006 2007
fund name sum MV # MV sum MV # MV sum MV # MV
AXA IMMOSELECT 242,312,000 9 388,679,000 10 CS EUROREAL (A) EUR 1,659,120,000 39 DEGI EUROPA 1,248,560,000 16 Deka ImmobilienEuropa 1,609,620,000 27 1,584,349,000 28 1,348,370,000 27 Deka-ImmobilienGlobal 129,790,000 2 Deka-ImmoFonds 4,095,137,091 129 3,457,142,739 126 3,159,100,000 115 EURO ImmoProfil 1,595,280,000 71 grundbesitz europa 4,017,193,000 98 2,468,795,000 66 316,089,000 3 HANSA Immobilia 171,385,000 19 HAUSINVEST europa 1,485,208,000 26 1,522,335,000 25 INTER ImmoProfil 338,680,000 13 MORGAN STANLEY P2 VALUE 68,480,000 2 SEB ImmoInvest 1,855,100,000 62 1,458,300,000 38 TMW Immobilien Weltfonds 90,880,000 2 UBS (D) Euroinvest Immobilien 623,700,000 22 437,500,000 15 UniImmo: Deutschland 3,647,410,000 57 1,967,550,000 22 UniImmo: Europa 1,715,310,000 25 UniImmo: Global 242,881,000 2 363,015,000 3 WestInvest 1 1,812,495,100 71 1,815,168,000 72 1,552,138,000 52 WestInvest InterSelect 442,226,000 11 612,369,000 14 816,655,000 15 total 12,106,461,191 338 18,034,434,739 484 20,217,426,000 512 min. 129,790,000 2 242,312,000 2 68,480,000 2 max. 4,095,137,091 129 3,647,410,000 126 3,159,100,000 115
8Sebastian GläsnerERES Conference 2009
Negative value changes and high dispersion in 2006
D MV analysis year 2005 2006 2007 # properties 338 484 512
mean -4.287 -8.411 -1.716
standard deviation 6.156 13.129 5.752
variance 37.895 172.381 33.091
skewness -1.595 -2.038 -1.396
kurtosis 6.806 5.764 7.697
9Sebastian GläsnerERES Conference 2009
High valuation disparity among funds
2005 2006 2007
fund name # mean sd skew. kurt. # mean sd skew. kurt. # mean sd skew. kurt.
AXA IMMOSELECT 5 0.045 0.025 0.053 -1.566 5 0.016 0.045 -0.212 -2.168 CS EUROREAL (A) EUR 26 -0.017 0.067 1.593 5.529 DEGI EUROPA 9 -0.037 0.145 -0.090 -0.748 Deka-ImmobilienEuropa 16 -0.051 0.048 -1.419 0.978 22 -0.121 0.099 -0.943 1.061 19 -0.036 0.068 -1.378 3.055 Deka-ImmobilienGlobal 1 0 Deka-ImmoFonds 97 -0.066 0.073 -1.205 5.749 117 -0.194 0.185 -0.749 0.943 52 -0.037 0.079 -1.307 4.011 EURO ImmoProfil 44 -0.031 0.056 0.014 1.251 grundbesitz europa 83 -0.047 0.059 0.110 0.590 65 -0.113 0.073 0.124 -0.290 0 HANSA Immobilia 15 -0.035 0.058 -0.106 1.010 HAUSINVEST europa 19 -0.109 0.162 -1.964 3.027 20 -0.017 0.075 0.022 1.091 INTER ImmoProfil 2 MORGAN STANLEY P2 VALUE 1 SEB ImmoInvest 34 -0.054 0.065 -2.093 5.410 23 -0.021 0.039 -0.907 1.125 TMW Immobilien Weltfonds 2 UBS (D) Euroinvest Immobilien 16 0.002 0.067 0.190 -0.785 9 -0.003 0.057 0.804 -0.978 UniImmo: Deutschland 26 -0.076 0.109 -2.154 5.534 13 -0.045 0.105 -0.285 0.728 UniImmo: Europa 12 -0.040 0.105 -0.504 -1.007 UniImmo: Global 1 WestInvest 1 47 -0.058 0.068 -2.136 7.463 49 -0.041 0.049 -1.087 1.149 38 -0.044 0.075 -0.968 0.758 WestInvest InterSelect 7 -0.038 0.065 -1.279 0.191 11 -0.027 0.048 0.019 -0.820 10 -0.012 0.040 -0.391 -1.081 all properties 338 -4.287 6.156 -1.170 5.509 484 -8.411 13.129 -1.652 4.095 512 -1.716 5.752 -0.689 3.441
10Sebastian GläsnerERES Conference 2009
Significant valuation differences among fundsTests of Between-Subjects Effects for 2006
Dependent Variable: D MV
Source
Type III Sum of
Squares df Mean Square F Sig.
Partial Eta
Squared
Corrected Model 20909.3 10 2090.9 15.862 .000 .251
Intercept 3505.3 1 3505.1 26.592 .000 .053
fundname 20909.3 10 2090.9 15.862 .000 .251
Error 62350.6 473 131.819
Total 117500.6 484
Corrected Total 83259.9 483
R Squared = .251 (Adjusted R Squared = .235)
>>> The research hypothesis „All means of value changes of German properties over all funds are equal” is rejected
11Sebastian GläsnerERES Conference 2009
Descriptive statistics of regression model
The model includes three factors to explain the changes in market values (D.MV):
year.of.purchase
occupancy rate
percentage ending leases
The factor change in occupancy rate (D.OCR) is not included due to data limitations.
2005 mean sd n NA year.of.purchase 1993.2 8.6 338 0 occupancy.rate 0.821 0.243 251 87 D.OCR -0.019 0.145 97 241 perc.ending.leases 0.142 0.248 251 87 YP 14.98 2.45 337 1 D.YP -0.010 0.044 336 2 D.ERV -0.034 0.049 336 2 D.MV -0.043 0.062 338 0 2006 mean sd n NA year.of.purchase 1993.9 9.0 484 0 occupancy.rate 0.831 0.241 474 10 D.OCR -0.009 0.152 378 106 perc.ending.leases 0.128 0.219 379 105 YP 14.50 2.86 471 13 D.YP -0.039 0.097 471 13 D.ERV -0.051 0.084 471 13 D.MV -0.084 0.131 484 0 2007 mean sd n NA year.of.purchase 1995.1 8.7 512 0 occupancy.rate 0.843 0.243 442 70 D.OCR -0.011 0.139 428 84 perc.ending.leases 0.141 0.198 284 228 YP 14.94 2.74 496 16 D.YP -0.004 0.040 494 18 D.ERV -0.013 0.046 494 18 D.MV -0.017 0.058 512 0
12Sebastian GläsnerERES Conference 2009
Significant valuation differences among funds (I) model variables
fund p-
value DF adjusted R
squared occupancy.
rate year.of. purchase
perc.ending. leases
Deka-ImmobilienEuropa 0.052 24 0.427 0.207 0.017 0.098 (2.050) (1.543) (0.245) Deka-ImmoFonds 0.000 113 0.374 0.361 0.001 -0.148 (4.949) (0.751) (-2.160) grundbesitz europa 0.008 62 0.243 0.168 -0.001 -0.005 (2.942) (-0.937) (-0.067) HAUSINVEST europa 0.397 11 0.467 0.172 0.004 -0.367 (0.700) (0.571) (-0.660) SEB ImmoInvest 0.152 17 0.351 0.097 -0.001 -0.073 (1.457) (-0.653) (-0.614) UBS (D) Euroinvest Immobilien 0.094 17 0.149 0.110 -0.016 -0.003 (2.305) (-2.272) (-0.052) UniImmo: Deutschland 0.149 53 0.218 0.035 0.001 -0.151 (0.465) (1.472) (-1.273) WestInvest 1 0.012 35 0.059 0.011 0.001 -0.048 (0.131) (0.466) (-1.602) WestInvest InterSelect 0.532 5 -0.191 0.023 -0.010 -0.063 (0.108) (-0.513) (-1.411)
The table reports results of the linear regression D.MV ~ occupancy.rate + year.of.purchase + perc.ending.leases. The three colums to the left report p-values, degrees of freedom and adjusted R squared values of the model, while the three colums to the right state the estimates of the independent variables. Figures in parentesis under the corresponding estimates report the t-values. Bold coefficients imply a p-value of 0.1 or lower
13Sebastian GläsnerERES Conference 2009
Significant valuation differences among funds (II) model variables
fund p-
value DF adjusted R
squared occupancy.
rate year.of.
purchase perc.ending.
leases AXA IMMOSELECT 0.001 7 0.891 -0.024 (-5.304) Deka-ImmobilienEuropa 0.011 25 0.436 0.181 0.018 (2.668) (1.671) Deka-ImmoFonds 0.000 113 0.374 0.361 0.001 -0.148 (4.949) (0.751) (-2.160) grundbesitz europa 0.000 64 0.255 0.166 (3.941) HAUSINVEST europa 0.072 24 0.388 0.413 (1.880 SEB ImmoInvest 0.002 60 0.106 0.081 (3.159) UBS (D) Euroinvest Immobilien 0.044 18 0.197 0.111 -0.016 (2.274) (-2.295) UniImmo: Deutschland 0.145 55 0.227 -0.164 (-1.478) WestInvest 1 0.002 69 0.196 0.061 0.003 (1.457) (1.995) WestInvest InterSelect 0.953 11 -0.159 0.026 -0.005 (0.215) (-0.287)
The table reports results of the linear regression D.MV ~ occupancy.rate + year.of.purchase + perc.ending.leases for those funds for which all three variables are significant. Not significant variables have been excluded from the model. The three colums to the left report p-values, degrees of freedom and adjusted R squared values of the model, while the three colums to the right state the estimates of the independent variables. Figures in parentesis under the corresponding estimates report the t-values. Bold coefficients imply a p-value of 0.1 or lower
14Sebastian GläsnerERES Conference 2009
Summary (I)
significant differences of property valuation between funds’ portfolios especially in 2006
likely explanation for different mean value changes of properties by portfolio:funds exerted an influence on the valuation of their properties
alternative interpretation: not fully diversified portfolioshigh vacancy rates explain devaluations in many (not all)
portfoliosThe fact that the same model is far less significant and the
explained variance is lower in 2005 and 2007 strengthens the argument that in fund crisis year 2006 many German properties were prepared for sale at (adjusted) market values
15Sebastian GläsnerERES Conference 2009
Summary (II)
big valuation differences by portfolio holding question research and indices neglecting this variable
Consequences for the idiosyncratic risk of property portfolios: if a big proportion of property value change variance stems from portfolio belonging and not from property-specific factors, idiosyncratic risk of property portfolios may be overestimated
16Sebastian GläsnerERES Conference 2009
Thank You