Jean-Paul Rodrigue (2013), New York:
Routledge, 416 pages.
Author: Dr. Jean-Paul Rodrigue
The Lowry model was one of the first transportation / land use model
to be developed in 1964 for the Pittsburgh region. Even if its formulation
is rather simple, it depicts well the relationships between transportation
and land use. Its premises were expended by several other models, known
as "Lowry-type" models. The core assumption of the Lowry model assumes
that regional and urban growth (or decline) is a function of
the expansion (or contraction) of the basic sector. This employment
is in turn having impacts on the employment of two other sectors, retail
Basic sector. Employment that meets non-local demand. It
produces good and services, which are exported outside the urban
area. It generates a centripetal flow of capital into the city generating
growth and surpluses. Most industrial sector employment is within
this category. It is generally assumed that this sector is less
constrained by urban location problems since the local market is
not the main concern. This consideration is an exogenous element
of the Lowry model and must be given.
Retail sector (non-basic sector). This employment meets the
local demand. It does not export any finished goods and services
and use the region as its main market area. It accounts mostly for
services such as retailing, food and construction. Since this sector
strictly serves the local / regional demand, location is an important
concern. Employment levels are also assumed to be linked with the
local population. This consideration is an endogenous element
of the Lowry model.
Residential sector. The number of residents is related to
to the number of basic and retail jobs available. The choice of
a residential area is also closely linked to the place of work.
This consideration is an endogenous element of the Lowry
Employment in the basic sector influences the spatial distribution
of the population and of service employment. This level of influence
is related to transport costs, or the friction of distance. The
higher the friction of distance, the closer places of employment (basic
and non-basic) and residential areas are. Overall, the Lowry model has
2. Data Process Structure
The model aims to establish a representation of the residential structure,
of employment and of services in an urban area. With an exogenous
spatial distribution of the basic sector employment and a set of
transport costs between zones, the model calculates total population
and employment by zone. It is composed of an
economic sub-model and a
spatial allocation sub-model, which are
subject to constraints.
- The residential sector, and thus urban land use, is a function
of employment. This function is calculated assuming multiplier
effects of basic and non-basic employment. Each job is thus linked
to a number of people.
- The total employment is a function of the employment in the
basic sector. The retail employment is thus the result of a multiplier
effects on the basic sector.
- The location of the population is a function of the costs involved
to go to their place of work, a gravity-based friction of distance
The two sub-models require a set of basic data and resolve the problem
- The first sub-model establish the impacts of the basic employment
over the non-basic employment and over the population.
- The second sub-model establish the distribution of the population
in function of attractivity and transport costs. This is done by
a gravity-type spatial interaction model.
The model can be singly constrained, that is the only constraint
is the fixed location of basic employment, as it is the case for the
below set of equations. It can also be doubly constrained, where
the location of basic employment and housing are fixed. The singly constrained
Lowry model is solved according to these equations:
- The spatial distribution of basic employment is assumed as given.
- The location of the basic workers is determined according to
a location-probability matrix, itself the result of a least friction
of distance function.
- Calculation of the residential sector per zone according to
the population per worker multiplier.
- Calculation of the number of non-basic workers per zone to service
the population. This is the result of a non-basic worker per capita
- The location of non-basic workers is determined according to
a location-probability matrix.
- Revision of the total population according to the population
per worker multiplier.
- Calculation of the total number of workers and the total population.
This is the summation of the basic and non-basic employment and
of the basic and non-basic related population.
- The above processes (4 to 7) are repeated until a convergence
is reached, that is an optimization of the equation system of the
model following a set of constrains such as density.
The Lowry model has obviously several limitations. It is notably
a static model, which does not tell anything about the evolution
of the transportation / land use system. Furthermore, current economic
changes are in the service (non-basic) sectors, forming the foundation
of urban productivity and dynamics in many metropolitan areas. Under
such circumstances the model is likely the be inaccurate in the major
service-oriented metropolitan areas of today. A way to overcome this
issue is to consider some non-basic service employment as basic. The
Lowry model does not consider movements of freight in urban areas, which
are very significant and have impacts on the friction of distance.
- Tij = Interaction from residential zone i to work zone j (work-related
- Sij = Interaction from residential zone i to service zone j
- Pi = Total population of a zone i.
- Ei, EBi and ESi = Total employment, employment in the basic
(B) and service (S) sectors for zone i.
- dij = Euclidean distance between zone i and j (in km).
- Alpha = population over basic employment multiplier.
- Beta = service employment over population multiplier.
- Lambda: Friction factor for residential interactions.
- Micron: Friction factor for services interactions.
- WTTRij and WTTSij = Willingness to travel for Residential (R)
or Services (S) between zone i and j.
- LPRij and LPSij = Locational probability for Residential (R)
or Services (S) between zone i and j.
- Click here to download
an operational basic Lowry model (MS Excel format).