Lowry Land-Use Transport Model
- Introduced by
- Ira S. Lowry
- Year
- 1964
- Type
- Spatial-interaction-based land-use and activity allocation model of a metropolitan area
- Family
- Land-use / transport interaction models
What does it do?
The Lowry model is the foundational operational model of urban land use, allocating where people live and where services locate around a given pattern of basic employment using spatial-interaction (gravity) distribution.
Devised by Ira S. Lowry at the RAND Corporation in 1964 as 'A Model of Metropolis', it treats the city as a system in which basic jobs attract resident workers, those residents demand local services, and the resulting service jobs attract still more residents — a chain solved by iteration until the whole system balances. It launched the entire field of land-use and transport interaction modelling.
The intuition behind the method
Start with the jobs a city has for reasons beyond its own boundaries — a port, a factory, a government office — call these 'basic'. Those workers have to live somewhere, choosing homes near their jobs but trading off commuting distance, so population spreads around the employment in a gravity pattern. All those residents then need shops, schools, and clinics, which create service jobs located where the customers are. But service jobs are jobs too, so they pull in yet more residents, who need yet more services. The Lowry model just runs this loop — people follow jobs, services follow people — round and round until the numbers stop changing, and that settled pattern is its forecast of the city's land use.
History
Ira S. Lowry built the model while at the RAND Corporation, publishing 'A Model of Metropolis' as research memorandum RM-4035-RC in 1964, originally calibrated to the Pittsburgh region. It was the first operational, computer-run model to allocate urban land use endogenously from basic employment using gravity-based interaction. John Garin reformulated it in 1966 with a cleaner matrix algebra (the Lowry-Garin model), and through the late 1960s and 1970s it spawned a family of descendants — the Garin-Lowry, ITLUP/DRAM-EMPAL, and many regional planning models — establishing land-use/transport interaction modelling as a field. Although later challenged by random-utility and activity-based approaches, the Lowry logic remains foundational.
When to use it
Use the Lowry model when you need a transparent, theory-grounded allocation of population and service activity across a metropolitan area's zones, driven by a known or assumed pattern of basic employment and a travel-cost matrix — for strategic land-use forecasting, testing the spatial consequences of a major employment change, or as the backbone of a land-use/transport interaction study. It is well suited to sketch planning and to teaching the people-follow-jobs, services-follow-people logic. It is less appropriate when household and firm location decisions need micro-level behavioural or economic detail, when the basic/non-basic split is unclear, or when dynamic, disequilibrium adjustment over time matters; modern activity-based or land-use-economic models are then preferable.
Limitations and considerations
- Aggregate and deterministic, ignoring the heterogeneity of households and firms in location choice.
- Static equilibrium with no explicit time path, so it cannot represent gradual urban change or lags.
- Requires an externally fixed basic-employment pattern, which the model does not explain or forecast.
- Omits land and housing markets, prices, and supply constraints that shape real location decisions.
Source conditions do not replace independent scientific review or validation by an analysis engine.
Strengths
- Captures the core circular logic of cities — people follow jobs, services follow people — in a single coherent model.
- Operational and computationally light, solvable by simple iteration even with modest data.
- Requires only basic employment, a travel-cost matrix, and a few ratios as inputs.
- Founded the land-use/transport interaction tradition and remains an intuitive teaching and sketch-planning tool.
Applications
- Strategic forecasting of metropolitan population and service-activity distribution for planning.
- Testing the land-use impact of a major new basic employer, such as a port or industrial park.
- Providing the land-use side of integrated land-use and transport interaction model systems.
- Sketch planning and scenario comparison of alternative regional development patterns.
- Teaching the economic-base and spatial-interaction logic of urban structure to planners and geographers.
How it works
How it works
The model begins with basic-sector employment E^B specified exogenously for each zone — the export-oriented or otherwise externally determined jobs (manufacturing, ports, universities, government) whose location is treated as a given input rather than something the model explains. This basic employment is the engine that drives the entire allocation; everything else in the city is derived from it.
Formula
E^{B}_j \;\text{given exogenously for each zone } jWorkers are allocated to residential zones in proportion to each zone's accessibility to employment, using a gravity/spatial-interaction distribution: the population P_i housed in zone i is a multiple of the employment in every zone j, shared out according to an impedance function f(c_ij) of the commuting cost and normalized across competing residence zones. A population-to-employment ratio converts employed residents into total population.
Formula
P_i = \alpha \sum_j E_j\, \frac{f(c_{ij})}{\sum_k f(c_{kj})}Given the population now distributed across zones, retail and service (non-basic) employment is located where the customers are. Service jobs E^R in zone j are a share of the surrounding population, distributed by an impedance function g(c_ij) of the shopping/service travel cost and normalized across competing service centres, with population-to-service ratios fixing how many service jobs each resident supports.
Formula
E^{R}_j = \beta \sum_i P_i\, \frac{g(c_{ij})}{\sum_k g(c_{ik})}Because the new service jobs are themselves employment, they generate additional resident workers, who in turn demand more services. Total employment in each zone is basic plus retail, E_j = E^B_j + E^R_j, and the residential and service allocations are repeated in alternation; each pass adds a smaller increment until population and employment converge to a stable equilibrium that is the model's land-use solution.
Formula
E_j = E^{B}_j + E^{R}_j,\qquad \text{iterate } P \leftrightarrow E^{R} \text{ until convergence}Common pitfalls
- Misclassifying the basic versus non-basic employment split, which propagates error through the whole allocation.
- Using poorly calibrated impedance functions, so commuting and shopping decay parameters distort the gravity distribution.
- Stopping the iteration before genuine convergence, leaving an unbalanced and biased land-use pattern.
- Letting zones exceed plausible capacity because the basic specification omits land-supply and density constraints.
Frequently asked questions
What is the difference between basic and non-basic employment in the model?
Basic employment is the export- or externally-oriented activity whose location is treated as given — manufacturing, ports, universities, government — and which drives the model. Non-basic, or service, employment is the retail and local-service activity that exists to serve the resident population and is located by the model itself. The split is central: basic jobs are the exogenous input, and population and service jobs are derived from them, so misclassifying employment corrupts the whole allocation.
Why does the Lowry model need to iterate?
Because the system is circular. Basic jobs generate resident workers, those residents demand services, and the resulting service jobs are themselves employment that generates still more residents, who demand still more services. A single pass cannot capture this feedback, so the residential and service allocations are repeated in alternation; each round adds a diminishing increment until population and employment stop changing, at which point the model has reached the equilibrium that is its solution.
How does the Lowry model relate to gravity and spatial-interaction models?
The Lowry model is built on spatial-interaction (gravity) distribution as its allocation engine. Both the residential step (workers distributed around jobs) and the service step (service jobs distributed around population) use an impedance function of travel cost to share activity among competing zones — exactly the gravity logic of the spatial-interaction model. Lowry's contribution was to wire two such gravity allocations together in a feedback loop driven by basic employment, turning a flow model into a land-use allocation model.
Related methods
Sources
Lowry, I. S. (1964). A Model of Metropolis (RM-4035-RC). RAND Corporation, Santa Monica, CA.
Open publication referenceCopied.