Skip to content U.S. Department of Transportation/Federal Highway AdministrationU.S. Department of Transportation/Federal Highway Administration

Office of Planning, Environment, & Realty (HEP)
Planning · Environment · Real Estate

HEP Events Guidance Publications Glossary Awards Contacts

Toolbox for Regional Policy Analysis Report (2000)

Case Study: Albany, New York

Overview

Albany, NY Aerial View

This case study illustrates the application of a basic land use model to assess the affects of potential land use shifts on congestion. The model was developed quickly and with low data requirements compared to a large-scale land use-transportation model. The approach can be useful for small or medium size MPOs wishing to assess the impacts of feedback between transportation and development patterns.

The Capital District, a four-county region surrounding Albany, New York, has experienced dramatic growth in vehicle-miles of travel (VMT) in recent years, which has far outstripped population growth. One cause of this rapid growth in VMT has been the dispersal of population and employment. While congestion does not yet represent a major threat to the area, future traffic increases are projected to increase traffic delay greatly and to affect transit and freight movement negatively.

Participants in the area's recent "New Visions" long-range transportation planning process expressed concerns over these trends and, in particular, over the transportation impacts of current and projected development patterns. In response, the Capital District Transportation Committee (CDTC), the metropolitan planning organization (MPO) for the Albany region, developed and applied a land-use "pivot" model. The model was applied to test the impacts of transportation actions, tax policies, zoning changes, and urban reinvestment on regional settlement patterns and mobility.

The results of the land use analysis showed that congestion by itself was likely to have a relatively small impact on regional settlement patterns. They also showed, however, that policies to encourage urban reinvestment could have a significantly positive impact on the transportation system. Traffic congestion and delay could also be reduced by locating new development in areas where excess highway capacity exists, away from critical congestion corridors.

This case study illustrates how a basic land use model can be applied, with relatively little effort, to expand the range of policies and impacts that are considered in the transportation planning process.

Context

The Capital District of New York is a four-county region with a population of 872,000. The district includes the cities of Albany, Troy, Schenectady, and Saratoga Springs (Figure 1). Regional vehicle-miles of travel (VMT) growth is greatly outstripping population growth; between 1980 and 1990 peak hour VMT increased by 37 percent while regional population increased by only five percent.

Figure 1. New York's Capital District

Albany Fig. 1 thumbnail

Source: Capital District Transportation Commission.

Figure 2. Excess Delay with Year 2015 Travel

Albany Fig. 2 thumbnail

Source: Capital District Transportation Commission.

The Capital District Transportation Commission's (CDTC) New Visions long-range transportation plan was developed between 1993 and 1997. The Capital District Regional Planning Commission (CDRPC) developed baseline land use forecasts that were used as a starting point for this plan. In the context of regional trends toward dispersal, participants in the New Visions planning process recognized the importance of land use policies for transportation planning and felt that the strong influences of land use and transportation on each other should not be ignored. As part of the New Visions process, a Demographic, Land Use, and Growth Futures Task Force was formed. The task force constructed various land use and development scenarios to evaluate the regional impacts of different policy choices. These scenarios took the form of "what if" questions that allowed discussion of which development patterns might be more desirable than others.

CDTC developed its land use model to assist in framing these "what if" questions quantitatively. The model was designed to provide insight into questions such as:

CDTC realized that even if a desirable development pattern could be articulated, there would be no guarantee that it could be achieved. They further realized that a variety of other social and economic impacts, in addition to transportation mobility and land development, need to be considered when evaluating policy options. Nevertheless, participants in the New Visions process found the insights provided by the land use model to be useful in assessing future alternatives and in making policy recommendations

Methodology

Land Use Model Overview

The CDTC Land Use Model was developed from available Lowry-Garin models such as those represented in the Highway Land Use Forecasting Model developed by Alan Horowitz. These models distribute regional employment and population among analysis zones, based on characteristics of the zones such as accessibility. CDTC's model differs from standard land use models in three respects:

  1. It uses measures of developmental attractiveness for residential development by zone that are derived from a multiple linear regression analysis of residential growth patterns in the Capital District between 1970 and 1990.
  2. It operates as a "marginal" model, rather than an abstract equilibrium model. That is, the model fixes in place the vast majority of dwelling units and a large portion of existing employment. The model allocates only the region's growth and a small portion of existing households and employment. (The fraction of existing land use re-allocated is based on the assumed likelihood that dwelling units or employment will be replaced over the time period; a larger fraction is assumed for employment than for dwelling units.)
  3. It is calibrated to Capital District Regional Planning Commission (CDRPC) regional forecasts of households and employment for the year 2015. That is, assuming constant travel times, property taxes, and property values, the model assumes that CDRPC forecasts by zone are correct.

Procedure

The specific steps in the modeling procedure are as follows:

  1. For each traffic analysis zone, compute the residential development strength (RDS) as shown in Equation 1, based on:
  • Potential number of new households that could be built in the zone, given developable land remaining and expected dwelling unit density;
  • Median price of owner-occupied housing units in the zone, divided by the median price of all owner-occupied housing units;
  • Ratio of total property tax rate per $1,000 full value in the zone to the mean property tax rate per $1,000 for all zones; and
  • Number of households in the zone at the beginning of each increment.

Equation 1

Equation

Where:

HHPOTj = Potential number of new households that could be built in zone j.
PRICERATIOj = Ratio of median price of owner-occupied housing units in zone j, to the median price of all owner-occupied housing units.
TAXRATIOj = Ratio of total property tax rate per $1,000 full value in zone j to the mean property tax rate per $1,000 for all zones.
HHBEGINj = Number of households in zone j at the beginning of each increment.
a, b, c, d, and k = Coefficients developed from regression analysis of historical residential growth in the region.
  1. Select an increment of total future household growth to allocate.
  2. Allocate this growth to each zone j based on the following factors (see Equation 2):
  • Residential development strength of zone j;
  • Employment accessibility of zone j (a function of total employment in each zone i and travel time between zones i and j); and
  • A zone-specific residential calibration factor.

Equation 2

Equation

Where:

dHHij,n  = Number of households allocated to zone j for workers employed in zone i for increment n.
dHHtot,n  = Number of households available for allocation in increment n.
EMPi  = Fixed employment in zone i.
dEMPi,1…n-1  = Employment added to zone i in previous increments.
RDSj  = Residential Development Strength of zone j.
tij  = Peak hour travel time by primary mode between zone i and j.
b = Gravity model exponent for work-to-home and home-to-work trips.
HHCFi  = Calibration factor derived from the ratio of CDRPC forecast household growth for the zone to the modeled household growth for the zone.
  1. Select an increment of total future employment growth to allocate.
  2. (Equation 3) Allocate this employment growth to each zone j, based on:
  • Employment accessibility of zone j (a function of total employment in each zone i and travel time between zones i and j);
  • Residential accessibility of zone j (a function of total employment in each zone i and travel time between zones i and j); and
  • A zone-specific employment calibration factor.

Equation 3

Equation

Where:

dEMPij,n  = Employment allocated to zone j based on households in zone i and other employment in zone i, for increment n.
HHi  = Fixed households in zone i.
a = Gravity model exponent for home-to-other and other-to-home trips as calibrated in CDTC’s travel model.
s = Gravity model exponent for other-to-other trips.
EMPCFi  = Calibration factor derived from the ratio of CDRPC forecast employment growth for the zone to the modeled employment growth for the zone.
Other variables are as defined in Equation 2.
  1. Repeat the process from step 1 until all future household and employment growth has been allocated.

Data Requirements

The following data were used in developing the CDTC Land Use Model:

For the "residential development strength" factor, CDTC selected variables that it found to correlate well with historical development trends in the Capital District. The residential development strengths (RDS) equation could be developed in different forms and with different variables. Analysts should review the availability of local data; and should test the effectiveness of alternative variables at explaining local development patterns. Alternative forms of these variables, such as log-transformations, could also be tested.

Calibration

RDS coefficients: The coefficients for the RDS equation are based on multivariate linear regression of Capital District residential growth between 1970 and 1990.

Travel time impedances: Travel time impedances for accessibility factors are based on gravity model friction factors from the CDTC's travel model.

Zone-specific calibration factors: To calibrate the model to regional employment and household forecasts, the following procedure is used:

  1. Run the model to allocate forecast (2015) growth under base year (1990) travel times, property taxes, property values, and developable land.
  2. Determine household and employment calibration factors for each zone. This is done using the ratio of the regional forecast growth by zone to the model's predicted growth increment by zone. This approach to calibration can save a significant amount of time, since it is not necessary to calibrating each of the equations used in the land use model.
  3. Re-run the model under future alternative scenarios (travel times, tax rates, etc.), with the calibration factors incorporated into the household and employment allocation equations.

Level of Effort

CDTC estimates that approximately three person-months of staff time were required to develop, calibrate, and apply the model. As the model is implemented through a simple program, it is straightforward to apply once the data have been assembled and the coefficients determined. The regressions were performed using a statistical package and the model was written and compiled by CDTC.

Application

Analysis Approach

Once the land use model was developed and calibrated against baseline forecasts, CDTC took the following approach to analyzing alternative transportation and land use scenarios:

  1. Define future scenarios/conditions;
  2. Run the land use model under 2015 "committed" travel conditions to determine the change in growth distribution under congested conditions;
  3. Run the land use model under 2015 "committed" travel conditions, in combination with other policy scenarios, to determine the impacts of each scenario; and
  4. For each resulting 2015 land use scenario, re-run the travel model to identify the impacts on transportation conditions.

Scenarios Analyzed

Six regional scenarios were identified, representing different levels and locations of growth. These include:

  1. Baseline growth forecasts.
  2. "Traffic Congestion." Baseline growth was modified by the impacts of congestion. Zone-to-zone travel times in the land use model were changed to reflect forecast congestion levels, and growth was reallocated among zones accordingly.
  3. "Traffic Congestion plus Property Tax." In addition to modifying growth in response to congestion, a uniform regional property tax was imposed. This was done by representing the ratio of property taxes per $1,000 of full value assessment to the regional mean as 1.0 for every zone of the region.
  4. "Southern Crescent Scenario." A diversion of growth to a southern crescent of towns (compared to existing northward expansion trends) was modeled by increasing the allowable residential density in five towns in this area to five households per acre. This is a proxy for the impact of expanded water and sewer infrastructure in these towns.
  5. "Urban Reinvestment Scenario." Before calibration, the land use model predicted greater development in existing cities than shown in regional forecasts. This suggests that the forecasts are implicitly accounting for variables such as perceptions or reality of school quality, crime, suburban attractiveness, etc., that are not explicitly measured in the land use model. An urban reinvestment scenario was modeled by removing the model's calibration factors on a zonal basis, thereby allowing the model to put more development in cities. This assumes that aggressive policies would be implemented to eliminate perceived differences in attractiveness between cities and suburbs.
  6. Higher regional activity. The impacts of alternative regional employment projections, as well as the location of this additional employment growth, were tested.

Impacts Measured

The following impacts of each scenario were measured:

All of these impacts (except for the land use measures) have been established as performance measures within the New Visions process. This core set of performance measures has allowed CDTC to evaluate widely differing projects, including transit, intelligent transportation systems (ITS), land use, and roadway improvement, in common terms

Findings

Some of the findings resulting from the CDTC land use modeling exercise include:

Figure 3. Growth in Households, 1990 - 2015

Fig. 3 Growth in Households, 1990 - 2015

Source: Capital District Transportation Commission, 1995.

Figure 4. Growth in Employment, 1990 - 2015

Fig. 4 Growth in Employment, 1990 - 2015

Source: Capital District Transportation Commission, 1995.

Figure 5. Effects of Urban Reinvestment

Fig. 5 Effects of Urban Reinvestment

Based on its discussion of transportation and land use issues, the Demographic, Land Use, and Growth Futures Task Force developed a number of planning and investment principles for the region. It further recommended a number of actions consistent with these principles. Some of these recommendations included establishing an urban service area; providing funding and staff participation in community-based, corridor-level land use/transportation plans; and surveying industries to determine the impacts of transportation on location decisions. The results of the CDTC Land Use Model assisted the task force in identifying the potential impacts of many of its recommended actions and helped the task force make more informed decisions.

Conclusions

Strengths

While CDTC's land use pivot model does not have the range of capabilities of large-scale urban models such as MEPLAN, TRANUS, or UrbanSim, it can be used to evaluate certain land use impacts with a modest level of resources. The model itself consists simply of three equations that are calibrated using historical data on population, employment, taxation levels, and housing prices, as well as travel times obtained from the regional travel demand model. In most urban areas, it should be possible to assemble these data with relatively little difficulty.

The primary strengths of the model are its ease of use and calibration based on local historical data. NCHRP Report 423A, Land Use Impacts of Transportation: A Guidebook, notes that the approach "may prove useful as a sketch planning tool because it provides a qualified improvement over the traditional cross-sectional implementation" (p. 73). The approach of allocating only incremental growth, consistent with an exogenous baseline forecast, can save a significant amount of time in calibration. Also, it eliminates the need to explain differences between locally-determined forecasts and modeled forecasts to planners and policy-makers.

CDTC's modeling exercise was useful in assessing the effects of congestion on land use patterns, as well as the relative effects of alternative land use distributions on congestion and mobility-related performance measures. The results appear plausible and are consistent with the results of land use modeling exercises in other urban areas. The "marginal allocation" nature of the model simplifies the exercise and also adds an element of realism, in the sense that existing development is likely to remain in use for many years. The CDTC's method of summarizing impacts by area type (e.g., central city, inner suburb) provides a useful framework for policy analysis.

Limitations

The CDTC model shares a number of limitations that are common to all land use models, to varying degrees. For example:

In addition, the approach described here has some further limitations that should be noted:

A discussion of other strengths and limitations of various land use models can be found in NCHRP Report 423A. Despite these limitations and approximations, the CDTC model probably gives a reasonable "first-order" approximation of impacts. Furthermore, it appears that planners and policy-makers in the Albany region found the results both believable and useful in assessing the potential effects of "what-if" scenarios for the region.

References

Published References

Capital District Transportation Commission. Evaluation of the Transportation Impacts of Land Use and Development Scenarios. Albany, NY (October 1995).

Poorman, John (1993). Case Study in Integrating Transportation and Land Use Decisions. Prepared for the Lincoln Institute of Land Policy and the Fall 1993 Statewide Transit Conference. Internet: http://ntl.bts.gov/DOCS/ilu.html.

Contacts

Organization Person Phone
Capital District Transportation Commission Chris O'Neil (518) 458-2161
Updated: 04/27/2012
HEP Home Planning Environment Real Estate
Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000