Communities change. Often this is due to growth, but it could also be due to decline. With a region, some communities may experience growth pressures while other in same region may experience be experiencing disinvestment and decline. The problem for planning is how to estimate what the need will be, and then to allocate that need to the available land (both developed and undeveloped).
Allocation of forecasted land use need is not purely a technical exercise. The principle of environmental sustainability argues for making the greatest use of already developed land. Economic efficiency argues for using undeveloped land, since it has a lower initial cost for land assembly and preparation. Social equity argues both ways. On the one hand, the services and social networks are already in place in developed neighborhoods, making it less time-consuming and expensive for people of limited means to live in those areas. On the other hand, the lower costs for purchase, assembly and site preparation permits developers to offer space at more affordable prices on undeveloped land. While these policy matters are not irrelevant to land use forecasting, they are implied but not directly addressed in the technique developed here.
The issue, then, is to determine the total need for various types of land use so that need can then be allocated among parcels of land available in the community. The technique presented here is based on Arthur C. (Chris) Nelson’s Planner’s Estimating Guide (APA Press, 2004). While this tool follows the basic design of Nelson’s Guide, it is a simplified model which assumes many of the standards that are developed much more extensively in his book. The workbook is available here. Note that it is a workbook, composed of multiple linked spreadsheets (the tabs at the bottom of the Excel of template). To help you follow along with the discussion that follows, a workbook completed with sample data from the Guide is also provided.
The basic approach is to determine the past trends in the drivers of land use (residential and employment), and use “rules of thumb” to convert those trends into projected demand for land. Besides the demand for residential and employment land, additional rules of thumb will be used to project the demand for public space. Transportation space is factored into the land uses, either as gross land use required (for example, for residential streets) or as employment land use (for example, for a railroad depot) and is not allocated separately.
The drivers of residential land and employment land are treated independently. The classical theory presumes that employment drives demand for housing; and in the long run and over an entire region, that is probably a valid assumption. However, any given community may “specialize” in employment opportunities (for example, the central city) while another may be predominantly residential (for example, a second-ring suburb). In fact, classical theory (Kaiser, Godschalk & Chapin, 1995) argued that one first determined “basic” employment, determined the residential need to serve that, then determined the “local” employment needed to serve both the basic employment and residential need, and then determined the “secondary” residential need derived from the service (local) employment. In this model, the total residential demand and the total employment demand are determined separately, based on past trends (adjusted for anticipated changes in those trends, if needed).
Note that trend lines and ratios can (and do) change over time. While the template attempts to control for the most likely sources of change, the possibility remains that any particular rate or ratio may fail to capture the local dynamics. The analyst should be prepared to change the calculated rates or the ratios supplied, should professional judgment suggest a more appropriate substitute.
Elements of the Model
1. Baseline Data: The baseline data table provides the basic observations that drive the rest of the model. Ideally, the user will supply data for three preceding decades and projections for two decades out. The template is set to use data from the previous census plus projections for two decades out (a total of 30 years) to develop the rates that will be used throughout the model. These timelines, of course, can be adjusted as need (and data) require, by adjusting the formulas in the spreadsheet.
The baseline data provide information on population and employment. Total population is divided into year-round residents and people living in institutions and group quarters (since such people make different demands on a city’s structure), and the total number of permanent households. The spreadsheet uses this to calculate total population and average household size. These data can usually be obtained from Census data. The model also adds the average daily peak season hotel/motel residential count (obtained from local sources) to arrive at an estimate of peak population. While total population is used to calculate the demand for residential and employment land, peak population is used to estimate the demand for public facilities.
Employment data are divided into non-urban (agriculture and extractive industries) and urban. While both are used to project demand for residential land, only urban-related employment is used to project demand for employment land and public facilities. Among the urban-related employment, “TCU” stands for “Transportation/Communication/Utilities” and “Services” is net of “Group Care” and “Hotel” employment, which will be treated separately when calculating public facilities. By convention, employment data are rounded to the nearest 10 or 100 (depending on the size of the community), since they are estimates in any event.
For both population and employment data, the spreadsheet will then calculate the totals, the percent change, and the annual rate of change.
There are a number of data sets which must be provided by the user. They are indicated in the template by cells that are shaded in blue. Some of the blue cells in the template are blank. These are for data that describe the unique characteristics of the community being studied. Other blue cells already contain data. These data represent “rules of thumb” (many of them explained in some detail in Nelson’s book), and may be left as is for a first analysis. One may or may not need to adjust these standard estimates, depending on the local circumstances. The cells that are not shaded represent data that the spreadsheet calculates for you.
The workbook sheet for residential land use can be read “backwards”—from right to left. The far right column (column Q) delivers the “Gross Plan Acres Needed,” which is the endpoint of the residential analysis—in the end, how much land are we going to need for each of the various uses? Working backwards from there, gross acres needed is derived from the “Net Plan Acres,” plus an allowance for “lost” land that is inevitable in any subdivision of land—land lost to easements, road rights-of-way, “outlots” of unusable land due to slope or soils or other factors. This leakage is labeled “Gross Acre Adjustment Factor” and rule-of-thumb values are provided in the spreadsheet. Net planned acreage is the sum of “New Net Acres” and “Remaining Acres” (remaining acres are the net of Existing Acres minus Planned Conversion—residential land that, in the time being considered, is likely to be converted to non-residential uses). New net acres is the result of dividing the Total New Units Needed by the Planned New Unit Net Density (this may or may not be the same as current net densities—rule-of-thumb values are provided, but they can easily be modified depending on local circumstances). In turn, total new units are difference between Total Units Needed and Total Units Current, plus an allowance for Total Units Lost (to demolition, fire, etc.—units which will be replaced in the housing stock rather than converted to some other use). The total units needed is the Occupied Units Needed, adjusted by Vacancy Rate New (which may or may not be the same as the current vacancy rate). The spreadsheet provides a rule-of-thumb value for vacancy rates, which the user is free to adjust. Finally, the number of occupied units needed is the product of the Estimated Population New divided by the appropriate Household Size New (which, again, can be estimated by a rule-of-thumb value).
The workbook sheet for employment land use draws on the projected employment data entered in the baseline worksheet. In addition, it requires the current acreage devoted to the various employment land uses. All the rest of the input data are rule-of-thumb measures. As before, reasonable values for those rules-of-thumb are provided in the template, but the analyst should feel free to adjust them as needed to approximate local circumstances.
The worksheet begins by calculating the Gross Square Feet per Employee. This begins with an estimate of the Square Footage per Employee (typical values are provided in the template), scaled to the Efficiency with which space can be used (again, typical values are provided). This results in Net Square Feet per Employee, which is further scaled by the expected vacancy rate (typical values are provided in the template) to provide the gross square footage required per employee.
The worksheet then turns to converting space-per-employee into acres needed. Each type employment land use has its own FAR (floor-to-area ratio), which controls how many workers can be placed on an acre of land. This amount of space (“Gross Square Feet per Acre”), divided by the “Gross Square Feet per Employee,” results in the number of Employees per Net Acre. However, not all employees in an industrial sector work out of the “home office” (construction workers, for example, are mostly at building sites and some “office” workers share space or even work at home). So the projected number of employees must be adjusted by the In-Place Employee Percent (again, a rule-of-thumb value is provided in the template). The resulting Planned In-Place Employees are then allocated space based on the number of employees per net acre, resulting in Planned Acres Needed. Existing Acres for each land use (which must be provided by the user) is then subtracted from the planned need, to arrive at the New Acres Needed. But these, recall, are net acres—they do not include the “lost” land discussed in the section above on residential land use. So the new acres needed are adjusted by a Gross Acre Adjustment Factor, to arrive at the Gross Acres Needed.
Nelson points out that the trend in the last half of the 20th Century has been to significant increases in the amount of space needed per employee. For example, in 1942 a typical office worker needed 110 square feet; in 2000, it was 280 square feet. In 1961, a typical manufacturing plant needed 389 square feet per employee; in 2000 it was 546 square feet. Nonetheless, the employment land needed per employee tends to be fairly consistent across the country and fairly stable in recent times. Nelson suggests that there is little space to be saved in commercial land needed; and, while there might be some savings possible in industrial land, the sector has been declining of late anyhow. The one area that he recommends examining for saving land is parking—by his analysis (pp. 49-51), most communities are oversupplied with parking by about 1/3.
4. Public Facilities: In addition to land for residential and employment uses, space must also be allocated for public facilities.
Nelson admits that “(p)ublic facilities are perhaps the most idiosyncratic of all land uses to estimate” (p. 72). But nonetheless he develops an extensive model for projecting what they might be. He begins by determining “functional population.” Often, the official census count of the population is not particularly useful for determining demand on public services. Some people live and work in the community; but others only commute in to work, while others leave the community in the morning and don’t return until the evening after work. So Nelson develops a tool for estimating the amount of the total time in a week that people in the various land uses are present in the community—a sort of weighted measure of the “active” population. Second, he points out that, while there are a few public facilities that have “industry standards” for their provision (public schools often have mandated space requirements, for example), most do not. So he develops a set of “level of service” standards for a wide range of public facilities. He applies these to land needs based on functional population, using a process similar to the one he developed for residential land.
In this workbook, I have chosen to follow Nelson’s initial advice—the need for land for public facilities will be highly idiosyncratic for each community. Rather than attempt to derive those needs from various rules-of-thumb, the worksheet requires the user to provide both the current land in use and the staff’s best estimate of projected need, given the population and employment projections that drive the rest of the model. The worksheet then calculates how much new land will be needed, adjusts the net land use for “lost land” to arrive at the gross acreage needed for each type of facility.
This worksheet requires almost no input from the user, deriving most of its data from the other worksheets. There is, however, one column, the Market Factor Adjustment, which the user may choose to adjust. Nelson points out that, once the total acreage for the planned need has been determined, that total should be adjusted upward to provide some flexibility in the market. Otherwise, as the community nears the end of the planned time horizon there will be almost no land available; this would provide windfall profits to landowners who monopolize what little remaining developable land remains, and would force others to develop marginal land that would be better left as it is. He recommends a planned surplus of 15-30% (less for larger communities, more for smaller ones—15% of a large number still provides a fair bit of flexibility in the market, while 30% in smaller markets may still not afford many usable parcels). He also recommends reducing it to 10 % for many public facilities (since they tend to work on longer timelines and have the advantage of eminent domain if needed).
So, in the end, the analysis delivers both a single gross acreage estimate and gross acreage estimates for each of the various categories of residential, employment, and public facility land uses. As Nelson warns throughout his book, these gross estimates do not necessarily provide any guidance on where the acres should be supplied; that is the substance of courses on urban design, subdivision regulation, and zoning. Nelson also points out that one may observe a significant impact on the land that is consumed by using only very marginal increases in density (for example, by increasing the standard from 3 to 4 units per acre). This could have a substantial impact on farm- and open-land conversion, and support more sustainable development practices.
The Planner’s Estimating Guide also includes sections on using land-use projections to estimate capital facilities costs, and a section on estimating the impact of unanticipated development. The workbook presented here provides a very much simplified version of what Arthur Nelson has created. The purpose of this workbook is to permit a quicker, less data-demanding estimate and to permit the student to understand the basic structure of land use projection. One is always encouraged to “go the source” for a more detailed analysis when time and funding permits.
© 1996 A.J.Filipovitch
Revised 21 July 2005