Geography Reference
In-Depth Information
Box 42.2
Estimating new store turnovers
Imagine you are a marketing analyst for a major car
company, say Fiat. You have identified Blackpool in
Lancashire, UK, as a large urban area without a major
Fiat dealer. Having built and run a spatial interaction
model (as given in equations 42.1-42.2), you now need
to analyse the results. The results below give the
predicted sales of your new Fiat dealer in Blackpool,
along with the sales lost from other Fiat dealers nearby:
With profit levels in the UK car industry at approximately
17 per cent for the manufacturers, and the average price
around £12,000, then the net gain in profit alone to Fiat
is 280× 12,000×0.17=£571,200 (assuming land costs,
etc. are not included in operating profits at this stage).
This example emphasises the importance of
modelling techniques that deal with all competitor outlets.
The prediction for the new site must be offset by the
predicted loss of sales from surrounding outlets. Clearly,
the company must examine these costs and benefits
(and also examine sales gained from the competition: in
this case, other Blackpool dealers of other major
manufacturers).
In the long term, a company may be interested to
know what the optimal locations for its local
network should be, given the objectives of
maximising either total sales or market share, and
how this compares with the existing distribution
network. Formally, the spatial interaction model
can be rewritten as a mathemtical programming
formulation. The following may be a typical
objective and set of constraints (given here in
words rather than equations!):
Maximise: Market share in region X for
organisationY
Subject to: Maximum number of stores
Minimum number of stores
Mininum store sales of £Y
Minimum inter-store drive time of
T
minutes
No consumer to be more than
M
minutes drive
time from a store
The problem can be solved either with existing
outlets
in situ
or with all outlets free to relocate (at
least theoretically). A heuristic algorithm has been
developed that solves this complex problem on a
PC (for a full description of the detail, see Birkin
et al
. 1995). Clarke and Clarke (1995) describe an
application of such optimisation techniques by
GMAP for Toyota in the US car market. These
models were asked to find two new locations for
Toyota in Seattle/Tacoma such that the impact on
existing Toyota dealers was minimised. The two
new locations produced a net gain in sales to
Toyota of 1000 units. Given an average
profitability of $1500 per unit, the overall increase
in profits in Seattle/Tacoma alone is of the order
of $1.5 million.
CONCLUSIONS
It is interesting to speculate on the future of store
location research. Some authors are convinced that
the increasing saturation of many retail markets (in
theory meaning no more new constructions are
possible or feasible) means that store location
research will become redundant. Clarkson
et al
.
(1996: p. 31) suggest:
As the UK grocery market becomes
increasingly saturated, the development of new
stores on new sites would diminish in
importance. The need for more sophisticated
location assessment procedures would then
become significantly less important to retailers
in their pursuit of growth strategies.
However, there are two major responses to such an
argument. First, in many markets the notion of
saturation can be challenged, even in the
sophisticated markets of the grocery sector in the
late 1990s (see Langston
et al
. 1997; 1998; Guy
1996). Second, it could be argued that the
increasing sophistication of retailing may result in a
greater need for store location research rather than
