Geography Reference
In-Depth Information
Figure 44.2
Cardiff's inner
area.
These principles were put into practice in the
following way. A survey of estate agents was
undertaken to identify the asking prices of
properties for sale at the time of the study. This
house price survey identified the asking prices of
796 properties (2.1 per cent of all properties in
the study area—296 of them identified as
individual properties, the remainder identified to
the street level of resolution), which were spatially
distributed even across the study area. On the
assumption that asking prices bore a
correspondence with actual values, a geographical
model was developed in order to allow
computation of the capital values of all properties
that were not actually on sale at the same time.
The stages in this model are summarised in
Figure 44.3. Essentially, the rates register was used
as the address frame, although quite a large number
of its entries required aggregation from
constituent rooms and bed-sits into whole
properties. The first stage entailed identifying the
precise geographical locations of the 269 'price
survey properties'. Next, those properties that
were in the same street and were of the same type
as a price survey property were allocated the same
capital value—or an average value if there were
two or more price survey properties in a given
street. This allowed capital values to be computed
for 20,982 properties. In the third stage, the
digitised house condition survey boundaries were
used to assign values to properties located in streets
in which there was no price survey property.
However, at this scale it was considered that 'street
effects', even of physically identical properties,
could conspire to invalidate use of values from a
neighbouring street. Consequently, the rateable
value of the nearest price survey property was
compared with the distribution of rateable values
in the adjoining street in order to ascertain
whether the value of a price survey property could
legitimately be used to compute the values of
some or all of its more distant neighbours—with
the acceptance criterion based on a standard
deviation measure. This allowed values to be
computed for a further 22,523 properties. This left
just 1884 of the 45,658 without assigned capital
values. These were overwhelmingly idiosyncratic
properties, so values were assigned using a simple
regression relationship between the asking prices
(denoted CVAL) and the rateable values (denoted
RVAL) of the price survey properties. This
empirical relationship was statistically estimated as:
CVAL=4003+358.64 (RVAL)
Housing exhibits a very wide range of
characteristics in terms of structure, age and
condition. Additionally, each individual property
