University
of Washington
Geography
367
Professor
Harrington
Public-sector uses of economic GIS
Contents:
The examples for these notes were
drawn from Chapter 7, "Model-Based GIS for Urban Planning," in Intelligent
GIS: Location Decisions and Strategic Planning, by M. Birkin,
G. Clarke, M. Clarke, and A. Wilson (Cambridge, England: GeoInformation
International; New York: John Wiley & Sons).
ASSESSING LOCAL-AREA NEEDS
Geographic data indicating the needs of particular populations
are available with widely differing geographies:
-
Address information on substandard housing
-
School-zone information on proportions of subsidized lunches
-
Census block group information on household demographics
(number children, single-parent households) and income
-
ZIP code information on buying patterns
-
Types of morbidity by local-clinic service area
GIS allows these differing geographies to be related to one
another, assigning values to some particular geographic unit (e.g., ward,
election district, service area) at which a program is to be administered.
This requires
-
manipulation of geographic data across different types of
zones (geocoding, cookie cutting), and
-
manipulation of different sorts of measures.
The latter is often accomplished by establishing ranked categories
for each measure, and then combining the measures in weighted or unweighted
fashion to yield aggregate measures of local need. In class, we discussed
two alternative approaches to creating a single indicator from a number
of measures. One alternative (which we've pursued in our cases) loses
even more of the information in our original data, and one alternative
uses more of the information in our original data. What
are these two alternative approaches?
RETAIL SITE ANALYSIS
The methods we learned in the previous
section of the course can be applied by public entities to determine
the impact of a new shopping center for which permits are sought. Based
on the attractiveness of the proposed center, existing centers, and the
geo-economic configuration of the market area, would existing centers (which
pay taxes and soak up infrastructure) be forced into marginality by a new
center?
This is not often much of a concern in U.S. planning.
LABOR AND TRANSPORTATION PLANNING
In class, I showed a map of the expected spatial distribution
of journey-to-work origins for a proposed new employment center ("science
park," what we tend to call "technology park"), of 100 employees. This
distribution is generated by using a spatial interaction model, in which
the relevant population of each zone is the working population with the
occupational mix needed in the park. The attractiveness of the park is
its employment size, and the distance measures are average drive times
from the centers of the zones.
This is useful for transportation planning and for assessing
the impact of a new employment center of unemployment rates by zone. (In
this case, however, 100 employees create a pretty small transport or employment
impact).
SITING PUBLIC SERVICE CENTERS
Approach A: opening new centers
In class, I showed a sequence of figures that illustrate
a process by which additional job-training centers were sited in central
Leeds. Geographic data were compiled to answer the following
questions:
-
Where do unemployed people live?
-
Where is there demand for job training, on the assumption
that unemployed people will or can travel only a specified, short distance
for training? The process used a spatial interaction model to identify
the demand for training in each zone of the city, recognizing
that people in one zone may be a source of demand for training in a nearby
zone.
-
What is the current level of availability of training, based
on the location and capacity of current centers, and the
distances that people can travel to get to them? (Again, this particular
process used a spatial interaction model, in recognition of the fact that
not all the demand for training at a particular center may lie within the
zone that the center is in).
-
Clients can be allocated to centers by minimizing the distance
that each zone's residents travel to a center, up to a capacity constraint
for each center (location-allocation) or
-
Clients can be allocated to centers according to a spatial
interaction model (SIM), in which Pi is the unemployed
population in zone i, Pj is the number of courses
or the capacity of courses at center j, and dij
is some measure of commute time.
-
Note that SIM formulas can be constrained to produce a specific
number of trips from all origins combined, or a specific number
of arrivals at all destinations combined, or a specific total of
origin departures and destination arrivals.
-
What zones have substantial deficits of demand in excess
of supply?
-
If a new center is added in south-central Leeds, how will
that change the supply configuration?
-
How will that affect the overall pattern of supply deficit?
Where?
Note that one could establish a continuous measure
of deficit, rather than a yes-no variable, and determine which sequence
of new centers would decrease the deficit most rapidly.
Approach B: determining which centers to enlarge or
subsidize
Alternatively, we could keep the existing centers, but
try to decide which should be improved.
-
Create a zone (buffer) around each center, based on maximum
desirable journey-to-center.
-
Define a set of measures that indicate need.
-
Compile these measures within each buffer area, using the
values for all zones (block groups, Census tracts, individual addresses)
that fall in part within the buffer.
-
Create a compiled variable of needs, and allocate resources
to the centers that have the neediest zones around them.
This is the core of Case 3 for this class.
What are some of the shortcomings
of this approach?
copyright James W. Harrington, Jr.
revised 9 February 2004