UrbanBuzz
Building sustainable communities
i-VALUL
iValul
4.2 KIT4
Residential property value
Prof Bill Hillier
UCL
Yolande Barnes
Savills
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Are there spatial correlates to residential
property values?
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.3 Knowledge Integration KIT4 – Residential property value
UrbanBuzz
Building sustainable communities
i-VALUL
iValul
•
This is an attempt to identify any underlying patterns there may be in the relation between property
values and street layout by using Council Tax Band as a proxy for property value. Council tax values
were of course originally based on property values, but real values have of course changed
considerably. But changes in absolute values will not matter if the underlying pattern of values has
remained similar, since it is the pattern of values that we consider here.
•
I use the database of 65453 residential buildings that was used for the crime work in the proof of
concept for crime and urban layout. In a sense, this work can be seen as an extension of the crime
work, since, if you recall, residential burglary rates were U-shaped in the borough, in that they were
higher for low and high council tax values and lower in the middle. So looking more closely at the
relation between council tax values and street layout should also clarify the features of the crime
pattern.
•
Practically speaking, I am continuing the statistical exploration of the database, but using council tax
values as the dependent variable rather than residential burglary or street robbery.
•
Can I say at the outset that I am as surprised by the results as I suspect you may be, and even more
I am surprised by their clarity.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.3 Knowledge Integration KIT4 – Residential property value
UrbanBuzz
Building sustainable communities
i-VALUL
iValul
Tax band
single
multiple
•
•
•
•
•
•
•
•
17
143
1518
18423
18599
5515
3115
215
158
2070
8693
3649
823
184
113
10
1
2
3
4
5
6
7
8
• 65453 residential buildings with 101849 dwellings
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.3 Knowledge Integration KIT4 – Residential property value
UrbanBuzz
Building sustainable communities
i-VALUL
iValul
Council Tax Bands
from red for high to
light blue for low.
Dark blue means
no residence
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Regression Plot
5.5
5.4
singlesNC/MDn
5.3
5.2
5.1
5
4.9
4.8
4.7
4.6
0
1
2
3
4
5
TAX BAND
Y = 4.502 + .114 * X; R^2 = .99
6
7
8
9
5000m
On the right is the pattern of global integration pattern for the 7000+ segments of Ludstown, plotted from red for high through
to blue for low. On the left the correlation between the mean integration of the tax bands and the tax bands rising form left to
right. The correlation is almost perfect with an r-square of .99. This means that the historic London principle through which
the most advantaged people occupied the most strategic streets still holds even though most of the area was built in the
nineteenth and twentieth centuries. This does not mean living on the main road, since as we saw there is little residence on
large parts of these. It is the next level down which seems critical.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Regression Plot
5.6
5.4
singleslogCH+2n
5.2
5
4.8
4.6
4.4
4.2
4
3.8
0
1
2
3
4
5
TAX BAND
Y = 4.116 + .154 * X; R^2 = .748
6
7
8
9
5000m
We find a similar, though less perfect effect for the choice or through-movement measure (sometimes called path
overlap).
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Regression Plot
9
8
TAX BAND
7
6
5
4
3
2
1
0
540 560 580 600 620 640 660 680 700 720 740 760
singlesNC/MDr2000met
Y = 1.388 + .005 * X; R^2 = .016
Regression Plot
9
8
TAX BAND
7
6
5
4
3
2
1
0
170
180
190
200
210
220
230
singles NC/MDr1000metric
Y = 13.417 - .043 * X; R^2 = .216
Regression Plot
240
250
9
8
TAX BAND
7
6
5000m
5
But as we reduce the radius of our integration or to-movement measure
to 2000, 1000 and 500 metres - so answering the question how
accessible are you from and to all points within so many metres real
walking distance ? – we find the effect gradually reverses.
4
3
2
1
0
50
55
60
65
70
75
80
singlesNC/MDr500metric
Mid Term Event
Y = 13.17 - .121 * X; R^2 = .581
85
90
i-VALUL the intangible value of urban layout
95
UrbanBuzz
© Space Syntax 2008
Regression Plot
singlesNC/MDr300metric
55
50
45
40
35
30
25
20
0
4
5
TAX BAND
Y = 49.303 - 3.035 * X; R^2 = .578
•
1
2
3
6
7
8
9
5000m
By 300 metres the effect is reversed. High tax properties have less local accessibility than low tax properties.
There is clearly a threshold above which high tax properties are well connected into the surrounding system and
below which they are not.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Regression Plot
4000
singlesTSL300metric
3800
3600
3400
3200
3000
2800
2600
2400
2200
0
1
2
3
4
5
6
TAX BAND
Y = 3738.214 - 181.631 * X; R^2 = .655
7
8
9
5000m
•
We find a very similar result by calculating the total street length available within 300 metres walking distance,
perhaps a more easy to intuit version of the measure.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Regression Plot
Regression Plot
20
1.8
18
1.6
sinlgesBUF1nonres
singlesBUF1res
16
14
12
10
8
1.2
1
.8
.6
.4
6
.2
4
0
0
4
5
TAX BAND
Y = 18.304 - 1.755 * X; R^2 = .879
•
1.4
1
2
3
6
7
8
9
0
1
2
3
4
5
TAX BAND
Y = 1.087 - .122 * X; R^2 = .425
6
7
8
We also find, not surprisingly, that tax band is inversely correlated with what we call building centred density –
the number of other dwellings wholly or in part within 30 metres of each dwelling. We also find ont the right that
higher tax bands also thin out the non-resdiential uses in their close vicinity.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
9
Regression Plot
Regression Plot
170
550
160
500
140
singlesLINElength
singlessegLENGTH
150
130
120
110
100
90
80
450
400
350
300
250
70
60
200
0
1
2
3
4
5
TAX BAND
Y = 52.722 + 14.072 * X; R^2 = .771
6
7
8
9
0
1
2
3
4
5
6
TAX BAND
Y = 235.179 + 34.238 * X; R^2 = .854
7
8
•
We also see – left above - that higher tax bands are associated with increased length of street segments
between junctions. This means that high tax properties tend to form part of larger urban blocks than low tax
properties.
•
Above right we see they are also associated with longer lines of sight along the street (without regard for the
number of junctions). You see a larger scale of urban environment from a high tax dwelling. Or, putting it the
other way round, high tax properties are more visually prominent in the urban environment.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
9
Univariate Line Chart
5.6
5.5
5.4
Variables
5.3
5.2
singlesNC/MDn
5.1
multipleNC/MDn
5
4.9
4.8
4.7
4.6
•
Observations
low to high
council tax band
All the results we have seen so far are for the 48000+ single dwellings i.e. houses. But it is also useful to
compare these with residential buildings with multiple dwellings. Here we compare the upward rise of the single
houses curve with the more complex curve for multiples. There may be two patterns for multiples, with the lower
tax bans dominated by social housing and the higher by private apartments or converted houses. But is it
notable that low tax multiples are substantially more integrated than single houses, but high tax multiples are
very similar to single houses.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Univariate Line Chart
6
5.8
5.6
Variables
5.4
5.2
singleslogCH+2n
5
multipleslogCH+2n
4.8
4.6
4.4
4.2
4
3.8
•
low toObservations
high council tax band
A similar pattern is found for the choice or through-movement potential measure.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Univariate Line Chart
6
5.8
5.6
Variables
5.4
singlesNC/MDn
5.2
singleslogCH+2n
5
4.8
multipleNC/MDn
4.6
multipleslogCH+2n
4.4
4.2
4
3.8
•
low toObservations
high council tax band
Here we see both types of dwelling on both variables. We must take care not to make too many inferences from
the extreme points since there the samples are much smaller – for example, we have only 10 H band flats for
the purpose peak top right. But these patterns are telling us a great deal about where we locate different kinds of
housing in the urban environment. Some social rule systems, as well as physical constraints, seem to be
UrbanBuzz
Mid operating
Term Event here.
© Space Syntax 2008
i-VALUL the intangible value of urban layout
Univariate Line Chart
25
22.5
20
Variables
17.5
singlesBUF1res
15
multiplesBUF1res
12.5
10
7.5
5
2.5
•
low to Observations
high council tax band
Here we see the reductions in building centred density for singles and multiples. Both densities fall with
increasing tax band, and are in fact very closely comparable.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Univariate Line Chart
5
4.5
4
Variables
3.5
3
sinlgesBUF1nonres
2.5
multiplesBUF1nonres
2
1.5
1
.5
0
•
Observations
low to
high council tax band
But if we compare buildings centred densities for non-residential uses, we find that in general multiples are much
closer to non-residential uses than houses, and there is a mark upturn for high tax bands. This will be much
affected by closeness to shops since this is the most common non-residential use in residential areas.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Univariate Line Chart
3.6
3.4
3.2
Variables
3
2.8
singlesSTOREYS
2.6
multiplesSTOREYS
2.4
2.2
2
1.8
•
low to Observations
high council tax band
We also find that – as we would expect – singles have more storeys with increasing tax band, and that is also
the tendency in the upper reaches of the tax band for multiples.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Univariate Line Chart
170
160
150
Variables
140
130
120
singlessegLENGTH
110
multiplesSEGlength
100
90
80
70
60
•
low to Observations
high council tax band
Here we see that singles and multiples both increase in the length of street segment with increasing tax bands,
and so are both part of larger blocks, though there is a fall in the higher tax bands for multiples.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Univariate Line Chart
550
500
Variables
450
400
singlesLINElength
mutliplesLINElength
350
300
250
200
•
low toObservations
high council tax band
Here we compare the visual prominence of singles and multiples. While singles increase more or less
consistently with higher tax band, multiples first increase strongly in the lower tax range, the fall and climb again
in the higher tax range.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
•
So we see that higher tax single properties:
•
- have greater accessibility to the large scale system, but less local accessibility
•
- have more passing larger scale movement passing the door, but less local movement
•
- are part of larger blocks
•
- have greater visual prominence in the ambient environment
•
- are farther from non-residential uses
•
- are higher buildings
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
•
For multiples we can say:
•
- multiples have higher accessibility than singles for low tax band and comparable for high
•
- multiples are closer to the shops and other non-residential activity than singles and this gets
stronger with higher tax band.
•
- densities for multiples are not too different from singles, and particularly close for higher tax bands.
•
- like singles, multiples become part of larger blocks with higher tax bands
•
- and gain in visual prominence with higher tax bands, though less consistently than single
dwellings.
•
Overall, we might say that the higher the tax band of the property the more it is oriented
towards the global rather than local system.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
Complex interrelationship between different factors
Univariate Line Chart
.18
.17
Variables
.16
singlesAGE
.15
multiplesAGE
.14
.13
.12
time periods from 1(early C19) to 7(late C20)
Questions:
Do the syntactic factors still affect Council Tax Band under
multi-variate analysis ?
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
Complex interrelationship between different factors:
Variables
Age
Property size
Ambient density
Layout
global factor
local factor
Multi-variate analysis
find out if syntactical factors are independent: Multi-variate analysis
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
First step: correlation matrices
Correlation Matrix
recipMD
TOmov2000m
TOmov500m
THRUmovCITYscale
THRUmov2000m
THRUmov500m
1.000
.649
.341
.531
.486
.344
TOmov2000m
.649
1.000
.624
.473
.501
.466
TOmov500m
.341
.624
1.000
.433
.466
.668
THRUmovCITYscale
.531
.473
.433
1.000
.946
.818
THRUmov2000m
.486
.501
.466
.946
1.000
.888
.668
.818
.888
1.000
recipMD
THRUmov500m
.344
.466
65453 observations w ere used in this computation.
One case w as omitted due to missing values.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
Stepwise regression of the
syntactical factors show:
ANOVA Table
TaxNum vs. 4 Independents
Step: 4
Split By: LUandRU=1then1else0
Cell: 1.000
DF Sum of Squares
Regression
Clustering (Integration) is much stronger
than path overlap (Choice)
Global integration (street prominence)
strongly and positively related
with higher tax bands
4
6816.283
Residual
48345
54832.863
Total
48349
61649.146
Variables In Model
TaxNum vs. 4 Independents
Step: 4
Split By: LUandRU=1then1else0
Cell: 1.000
Coefficient
Intercept
Local integration is strongly negatively
associated with higher tax band.
TOmovCITYscale(1/MD)
TOmov500m
THRUmovCITYscale
THRUmov500m
Mean Square
F-Value
P-Value
1704.071 1502.444
<.0001
1.134
Std. Error
Std. Coeff.
F-to-Remove
3.287
.045
3.287
5435.459
13.795
.349
.208
1558.955
-.010
2.083E-4
-.303
2340.856
.064
.006
.103
134.403
-.060
.014
-.043
18.533
Question
Do these effects survive, when we add:
age, property size and ambient density?
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
Question
Do these effects survive, when we add:
Variables In Model
TaxNum vs. 7 Independents
Step: 6
Split By: LUandRU=1then1else0
Cell: 1.000
Coefficient
Std. Error
Std. Coeff.
F-to-Remove
- age,
Intercept
-.309
.092
-.309
11.404
TOmovCITYscale(1/MD)
6.322
.288
.093
482.297
- property size and
TOmov500m
-.006
1.797E-4
-.185
1197.336
.036
.008
.025
22.546
- ambient density?
log(area*storeys/REScount)
2.781
.031
.391
7837.393
sqrtBUF1grndres
-.540
.009
-.247
3532.312
Age
-.045
.004
-.040
100.197
THRUmov500m
Variables Not In Model
TaxNum vs. 7 Independents
Step: 6
Split By: LUandRU=1then1else0
Cell: 1.000
Partial Cor.
THRUmovCITYscale
.007
F-to-Enter
2.345
Result
Both locational variables are weakened,
but remain very strong,
so their effects on Council Tax Band are to a
considerable degree independent of
size, density and scale factors.
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
In general
- Property size is by far the most important single factor in tax band
- Density is next – lower is higher tax
- Local spatial clustering is next – less means higher tax
- Global spatial clustering is next – more means higher tax
- Age is a positive, but relatively weak – older mean marginally higher tax
- Local path overlap level is weakly beneficial – more means higher tax
- Global path overlap level is immaterial
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
Importance of the factors
Variables
1. Property size
+
2. Density
-
3. Local clustering
-
4. Global clustering
+
5. Age
+
syntactical factors are
independent
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
4.2 Knowledge Integration KIT4 – Residential property value
Conclusion
So in spite of the clear inter-relations between our
syntactic/locational variables and age, property size and
ambient density, we can be confident that the effects we found
on council tax banding from the syntactic/locational variables
in the banding analysis are statistically independent effects –
at least in this borough.
Question
How does council tax band correlate to residential property value?
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008
Brent Value
Distribution
Brent Council Tax
Brent sales (Q206 to Q107)
35%
30%
25%
20%
15%
10%
5%
0%
<160,000
160200K
Source: Savills Research
200260K
260340K
340470K
470620K
Inflated Council Tax Band value
6201.2m
>1.2m
Brent Stock &
Transactions
% of Stock
% of Transactions Q107-Q407
60%
value growth q/q
50%
40%
30%
20%
10%
0%
Detached
Semi-detached
Source: ONS & Land Registry
Terraced
Flat
UrbanBuzz
Building sustainable communities
iValul
This project was supported by the UCL led
UrbanBuzz Programme, within which UEL is a
prime partner
presentation end
a.sdfhgsdl@spacesyntax.com
www.spacesyntax.com
Mid Term Event
i-VALUL the intangible value of urban layout
UrbanBuzz
© Space Syntax 2008