Development capacity and land budget modelling

Development capacity

The amount of human activity -- dwelling, working, and recreation -- that a given piece of land can accommodate (its development capacity) is determined by several interrelated factors:

  • Physical: Inhospitable terrain may make portions of the land unbuildable.
  • Infrastructural: The provision of water and sewer services may be limited by the capacity of local water bodies and watercourses.
  • Technical: Although materials technology and engineering techniques are always advancing, there are technical limits on what can be built.
  • Economic: While physical, infrastructural, and technical constraints may be overcome, certain building forms may be too expensive to construct under prevailing market conditions.
  • Social-cultural: While the other constraints may be overcome, the form and density of urban development or the mixture of particular uses may be constrained by cultural or social norms.

In practice, development capacity is codified in and regulated by development standards and building codes entrenched in plans, zoning by-laws, policies, and regulations, and informal rules-of-thumb. As discussed in Section 2.3, planning authorities commonly specify minimum and maximum levels for certain variables, including residential density, parking, and provision of public facilities. These can be expressed in terms of population, dwellings, land area, or distance. For example, municipal plans typically set requirements for parkland in proportion to the residential population and, to ensure access to them, define maximum catchment areas for particular sizes of parks. At the parcel scale, parking requirements for residential and commercial uses are typically determined on a per-resident basis or in proportion to floor area (De Chiara et al. 1995).

Types of land budgeting models

Much as fiscal budgeting involves the allocation of money to particular expenditures, land budgeting involves allocating land to particular uses or, put another way, determining how much land is required to accommodate anticipated future growth or change. There are two general types of land budgeting models: land-optimizing and activity-optimizing.

In a land-optimizing model, the objective is to determine how much land is needed to accommodate a given population. This approach is used in long-term planning, typically at the municipal or metropolitan region scale (e.g., see Kaiser et al. 1995: ch. 12). Land-optimizing models typically follow four steps. First, the future population of the area is forecast. Second, assumptions with respect to demographic change -- for example, the rate at which household size is declining -- inform a forecast of how the anticipated population will sort itself into households. Third, the household structure is translated into a profile of housing demand: the quantities of different types of dwellings require to house the forecast population. Finally, the land required for uses in proportion to population -- employment, schools, parks and other open space, hospitals, etc. -- is calculated.

Nelson's Planner's Estimating Guide (2002) is a land-optimizing model. For a given number of population and jobs, Nelson's model calculates the land required to accommodate housing, employment, and public facilities such as parks, schools, and religious institutions, as well as the amount and capital cost of necessary water and wastewater infrastructure. The Neptis Foundation's Toronto-Related Region Futures Study (IBI Group 2002, 2003) and the Government of Ontario's Projection Methodology Guidelines (MMAH 1995b) are further examples of land-optimizing models.

In an activity-optimizing model, the objective is to determine the optimal capacity of a fixed quantity of land -- i.e., how many people, jobs, and associated uses it can accommodate. This approach is often used by real estate developers who wish to estimate the potential capacity and economic yield of a known piece of land. Hosack's Land Development Calculations (2001) is an activity-optimizing model for parcel- or subdivision-scale development. This model allows for the intricate manipulation of built form elements within the parcel, including parking and loading requirements, land coverage of buildings, yards, and driveways, as well as the internal elements of buildings, such as the floor area dedicated to mechanical space and the number of floors.

Inputs and outputs

The number of input variables that can be incorporated into a model is potentially limitless. Accounting for every possible input variable, however, may not be desirable or necessary. A parsimonious model -- that is, one that requires the fewest possible inputs -- is more manageable for several reasons.

First, the input variables incorporated into the model should be appropriate to the geographical scale under study. At the municipal scale, the size and number of loading bays in a business park matter little; at the scale of the block, intra-parcel elements such as the sizes of garages, yards, and driveways matter greatly. Many employment, shopping, recreation, and education facilities serve a larger population than that of their immediate neighbourhoods and therefore operate at a broader scale than the site or subdivision. In addition, individual development sites exclude large-scale elements such as expressways, railway and electric power corridors, and protected natural heritage systems. For activity-optimizing models used at the scale of a single development site or subdivision, this means that amenities and large-scale infrastructure that serve a broader population need not be accounted for, because they are likely to be located elsewhere.

Second, input variables may interact with each other. The more they do so, the more likely it is that they are outcomes of a causally prior factor. In statistics, this phenomenon is known as collinearity. In this event, the prior factor is a more appropriate input variable.

Third, there is the acquisition of appropriate data. A model can only be made to work if necessary data and information are available and of high quality. The more parsimonious the model, the fewer the problems with data availability and consistency.