Essentials for Transit-Oriented Development Planning: Analysis of Non-work Activity Patterns and a Method for Predicting Success

By Dick Nelson and John Niles

Download this paper in Word 6.0


One hoped for benefit of transit-oriented development (TOD) is that a mix of shopping, service, and recreation activities at urban centers linked together by high quality transit will induce citizens to drive less and walk or ride transit more. Consequently, the success of the TOD concept depends greatly on the response of developers, consumers, and taxpayers to the new land use-transportation configuration.

Developers and owners of establishments that provide goods and services are expected to depart from highly independent siting criteria that now either result in large clusters of retail activity or cause stores to occupy stand alone sites, both with high levels of drive-to access. Consumers are expected to choose activities that are within close proximity rather than from among the great variety available in the regional marketplace. And taxpayers are expected to fund much greater levels of transit service needed to support the new land use patterns and induce significant numbers of new transit riders.

Efforts to predict the success of TOD would benefit from a much better understanding of nonwork activity patterns and trends on a metropolitan or travel corridor scale. Of the sixteen factors that will contribute to success on a regional scale, only a few are crucial at the local or station-area level, where more professional attention to success factors is observable.

A review of current evaluation methodologies that have been applied to TOD indicates that all have limitations. Travel demand modeling, in particular, cannot accurately predict the response of the marketplace to major transportation and land use changes. The authors outline a new approach to evaluating the likely success of TOD. This approach, termed Backcasting Delphi, is an exercise that pulls in critical thinking from key stakeholders, analysts, and planners. It starts with the desired TOD future and attempts to predict the circumstances needed for it to be realized, and whether those circumstances are achievable, given the forces shaping urban retail structure.


Transit-Oriented Development (TOD) has rapidly emerged as the central urban planning paradigm in the United States. Leaders in many metro areas have made, or are contemplating, major investments in new rail transit capacity, under the assumption that synergy between compact, mixed-use development and mass transit will change auto-dependent growth and travel patterns (1).

From an economic perspective, success of TOD will depend on the benefits -- both societal and personal -- it produces relative to its costs (Table 1). The public may experience benefits in the form of congestion reduction and air quality improvements. To the extent that TOD reduces excessive infrastructure costs associated with dispersed development, these would be accounted as secondary public benefits. The principal personal benefits may be travel time and expense saved, in addition to reduced congestion time. Personal benefits also include the possibility that some households can reduce the number of cars they own and operate. Other benefits, of a social nature and more difficult to quantify, may be associated with the enhanced quality of living TOD is believed to produce (2).

Public costs are primarily the transit capital and operating costs. The cost of housing in proximity to stations may be higher. Other direct costs may arise. To the extent that increased density does not result in reduced travel, congestion mitigation measures may be required. There may also be costs associated with TOD planning and any public incentives that may be needed.

TABLE 1 Simplified TOD cost-benefit accounting.



  • Transit system construction

  • Transit system operations

  • Mitigation of traffic congestion caused by compact development

  • Station-area housing cost premium

  • TOD planning

  • Public incentives to developers

  • Congestion reduction lowering time delays and fuel consumption

  • Air quality improvement reducing health costs

  • Reduced infrastructure

  • Personal travel time savings

  • Vehicle operation savings

  • Personal vehicle ownership reduction

In the context of planning, success of TOD depends on the response of developers, consumers, and taxpayers to the concept and to the public strategies that encourage it. Niles and Nelson (3) have identified 16 factors that will determine success at the regional or transit corridor level (Table 2). Fewer factors will control success at a single station-area, a main focus for planners to date.

TABLE 2 Factors determining the success of TOD.

Factor Station area success Regional success
Number and siting of TODs (station areas)   X
Transit quality   X
Transit technology   X
Street pattern X X
Station area parking X X
Employment and housing density X X
Commercial mix X X
Retail siting criteria   X
Regional market structure   X
Consumer activity patterns   X
Travel behavior/trip chaining   X
Zoning flexibility/land assembly X X
Resident reactions X X
Housing type preference/life style & life stage   X
Self-selection in residential choice X X
Government policies   X

The success of TOD is not without considerable obstacles (4). Growing wealth and technological innovation has lead to a very rich and diverse retail marketplace, which in turn generates increasing numbers of trips for nonwork purposes. Table 3 lists the retail trends that are readily observable in major US metro areas. Travel for nonwork purposes now constitutes 4 of 5 trips, and many commute trips involve stops for nonwork activities.

TABLE 3 Key retail trends.

Retail activity increasingly polycentric and dispersed
Planned shopping centers dominate market
Smaller malls cluster around major malls
"Big Boxes" market share growing
"Super" stores growing in kind and number
Many chains prefer stand alone sites
Dining out continues strong
Drive to and through convenience growing

These trends, the result of powerful forces operating in the consumer marketplace, constitute a significant challenge to TOD planners. Figure 1 attempts to describe schematically the situation. Stated as a question, the problem is: Can major transit investments, together with policies encouraging development near transit stations, sufficiently offset the behavior generated by expanding consumer income and preferences, as well as by market and technological innovation, to produce benefits commensurate with costs?

Figure 1:

Decision Tools for Determining the Success of TOD

In order to estimate the likelihood of success, planners are faced with determining the travel impacts of TOD on a regional scale when there is no existing example from which to learn. This is really a sub category of the general problem of estimating the effect of land use policies on travel patterns. Handy (5 ) has reviewed and critiqued alternative approaches used to explore the link between urban form and travel behavior, so we do not cover these efforts here. Instead, we look at methods that have been used to predict the success of TOD, and in particular, how well they deal with the strong effect of nonwork activities.

Predicting the success of TOD is made difficult by the large number of variables that must be taken into account. As the factors in Table 3 suggest, regional success of TOD is determined at three general levels: 1) the available public resources that limit the number of TOD stations and the quality of the transit service, 2) the response of developers and store owners to the market opportunities associated with TOD locations, and 3) the change in travel behavior of consumers in response to the activities available at TODs. Each interacts with the others. All, separate or together, must be amenable to analysis if the decision making process is to be adequately informed. And the ultimate challenge, when regional success is at question, is to choose tools that can assist the process even when there exists no operational regional TOD network to provide empirical data that establish the functional relationship between a factor and change in travel patterns.

Table 4 lists some of the decision-supporting methodologies that have been applied to TOD planning and the more general topic of New Urbanism in approximate order of their increasing complexity. Each of the methods has both advantages and limitations that will be briefly discussed, along with selected instances where each method has been applied. This list is not meant to be exhaustive of either the tools that are available or their real-world applications. The list is simply intended to show that a range of methods are available to help planners and decision makers ascertain the likely success of TOD.

TABLE 4 Examples of decision-support methods that have been applied to TOD and New Urbanism.

Method Selected application/reference
Structured discussion Berkeley Planning Journal (6)
Visual simulation Cervero & Bosselmann (7)
Metro area case study (interviews, field inspection) Boarnet & Compin (8)
Economic estimation Downs (9); Luscher (10)
Sketch planning Fox & Bowlby (11)
Travel demand modeling - assumed growth scenarios Thompson & Audirac (12)
Travel demand modeling - estimated real growth Puget Sound Regional Council (13)

Structured Discussion

Clearly the simplest method, structured discussion might merely be a roundtable discussion involving professionals, both practitioners and academics, who are involved in TOD and related issues. Although lacking in comprehensiveness and analytical component, it can yield useful insights, especially if the discussion is summarized and salient points are reported to the broader community of stakeholders. In 1994, Berkeley Planning Journal assembled eight planning professionals and scholars for a discussion of New Urbanism. Participants were asked to read at least two of four influential books and a recent critique. An abridged transcript was published by the Journal (6 ).

Visual Simulation

Since there are few contemporary examples of TODs that have textbook design features, it is difficult to gauge potential market demand, and this may be a factor in deterring TOD development activity. The concern is that consumers are reluctant to embrace what they have not experienced. To get around this obstacle, Cervero and Bosselmann (7) created photo images of "transit villages" with different densities and amenity mixes. They used these computer-generated simulations to query a representative sample of San Francisco Bay area residents as to their preferred neighborhood. Although an individual’s response to the look of a hypothetical neighborhood cannot possibly encompass all of the factors that determine actual home selection preference, it can suggest to developers and policy makers density and design alternatives that may not have been fully tested or may even be disallowed under existing land use regulations.

Case Studies

Even though TOD experience is limited, case studies can reveal factors that control the real-world form of TOD and the speed at which it can be implemented. Boarnet and Compin (8) used interviews and a review of zoning codes and planning documents to ascertain how localities in San Diego County planned for and implemented TOD. They were able to draw conclusions as to the role of local interests versus regional interests and the time frame for benefits to be realized, information that might be generalized to other urban areas considering TOD and rail systems. A case study will not determine whether TOD will succeed in other urban areas, but it can help identify general issues beyond the factors in Table 2 that may support or limit success.

Transit Cost and Performance Estimate

Downs (9) employed a simple but straight forward approach to determine the performance of TOD by estimating the number of TOD centers, assuming their density and size, that would be required to accommodate the growth of a hypothetical metropolitan region over a 10 year period. Downs used the average population and employment growth during the 1980’s of metropolitan areas that had achieved a 1990 population of one million or more. This allowed him to infer the number of transit stations and the financial feasibility of the regional transit system needed to support TOD. Luscher (10) used a similar approach to analyze the impact of TODs on auto travel in the San Francisco Bay area. He was able to estimate the reduction of regional per capita VMT TOD would produce, and to draw conclusions regarding its appropriateness as a regional congestion management strategy. Although such cursory methods can provide an important perspective on the tradeoff between transit system cost and transportation performance of TOD on a regional scale, they may not yield a clear indication whether success is possible at a corridor level, especially where there is opportunity for redevelopment and infill.

Sketch Modeling

In order to reduce the effort required for traditional demand modeling of TOD (see below), Bowlby and Fox (11) created a sketch planning tool. Essentially an abbreviated version of the regional demand model, the procedure allowed the transportation and air quality benefits of several TOD "packages" involving different levels of land use change, transit service improvement, and parking management for Memphis, Tennessee, to be evaluated. The tool permitted conclusions to be reached as to whether desired performance measures, including cost effectiveness measures, would be met. As discussed below, this method, while simplifying the analytic problems, necessarily suffers from the same difficulties all methods using regional travel demand forecasting models will encounter when applied to TOD.

Four-Step Modeling - Simulations

A similar but somewhat more complicated approach is to use a travel demand model for the simulation of regional or corridor TOD performance, providing land use inputs in the form of various plausible scenarios for residential and employment growth concentrations. Thompson and Audirac (12) did this for a Sacramento corridor, assuming that 105,00 more people and 52,000 more jobs would be distributed among urban TODs, neighborhood TODs, the central business district, and the urban fringe in differing amount. They estimated how transit mode choice and ridership for residents and commuters would be affected under each of four assumed scenarios. They were also able to draw conclusions about the most effective design of TODs, particularly the concentration of employment versus housing.

Four-Step Modeling - Real Growth Estimates

Metropolitan regions typically employ travel demand models to predict transportation system performance and air quality for expected land use scenarios. Land use changes are usually input assumptions to the models. In the case of the central Puget Sound region, planners assumed that a portion of new housing and employment will be concentrated in 21 designated urban centers as a result of regional growth management and transportation strategies (13). These are educated guesses by planners familiar with local development patterns and, at best, can provide rough estimates of the real transportation system performance of TOD as determined by market forces and government policies.

The limitations of traditional travel demand modeling in applications that involve forecasting the long-range travel changes resulting from nonwork activities associated with TOD are significant (14, 15). Data for locational attributes - prices, quality, and variety of goods at different shopping locations -- needed to specify consumer utility is lacking. Modeling studies of store location and consumer behavior dynamics have been few. And without previous experience with regional TOD networks, empirical data is scant that would allow calibration of models even if they were accurately specified. These same limitations will also hinder the application of activity-based modeling, which some believe will eventually replace four-step demand models.

Recommended TOD Decision-Informing Process: Backcasting Delphi

The evaluation methods just reviewed assist but do not offer full and convincing guidance on the difficult question of how to plan TOD on a regional or corridor scale such that the benefits justify the necessarily large investment in transit infrastructure and operations. There is a clear need for a new planning approach that more directly matches the difficulty of the problem. A new tool is required that takes into account the complexity of the metropolitan retail structure -- a structure that will continue to evolve dynamically. It should employ available descriptive data and information, yet it should not demand that only quantitative results be relied upon for estimating the likely impacts of TOD.

We believe a method that can meet these requirements is Backcasting Delphi, which melds the older approach of using diverse expert opinion to reach consensus through an iterative process (Delphi study) with the newer approach of working back from the desired outcome to determine the necessary policies and other inputs that will produce that outcome (Backcasting). See the following box for a description of Backcasting by one of it inventors.

Backcasting and Delphi have been applied separately and together to energy and transportation forecasting, principally in Europe and Canada. Hojer (16) used Backcasting Delphi to study the impacts of transport telematics under three passenger transportation scenarios: improved road system with user fees, improved public transit through rider information, and dual mode system which combines the flexibility of the private car with the capacity of public transport (private vehicles operate both on and off automatic guideways).


Backcasting’s concern is not with what futures are likely to happen, but with how desirable futures can be attained. It involves working backward from a particular desirable future endpoint to determine the physical feasibility of that future and what policy measures would be required to reach it. Targets are continually revised as new knowledge is acquired.

- Karl Dreborg, Essence of Backcasting, Futures 28(9), 1996, pp. 813-828.

As adapted and applied to TOD, Backcasting Delphi would utilize a multidisciplinary panel that includes urban planners, architects, urban geographers, urban economists, commercial developers, store site selection managers, transportation planners, and environmental organization representatives. Before undertaking the evaluation, they would specify the problem and establish the purposes, goals, objectives, boundaries, and other important parameters of the evaluation exercise. Table 5 briefly describes specific steps in the process that might then ensue.

TABLE 5 How Backcasting Delphi might be used to predict the metro-wide success of TOD.

Step Scope
Describe present retail structure/patterns Present urban structure including retail market, travel patterns, past trends
Identify forces shaping urban form Understanding and subjective weighting of forces: economic, environmental, social, technological. Focus on current and future market trends: commercial development, consumer behavior, nonwork travel patterns
Specify TOD scenario Likely station-area locations and types (residential, retail, employment, mixed)
Specify transit system Size and quality of transit afforded under fiscal constraints
Define success Economic, societal, personal, and environmental benefits and costs; elaborate 16 planning factors; establish planning horizon
Evaluate success Identification of constraints and supporting policies to achieve feasibility; adaption to new knowledge & consideration of alternative solutions as needed

For the purpose of TOD evaluation, Backcasting Delphi provides several advantages over other methods. In the ideal case, it would precede decisions to invest in transit capacity. It would allow involvement of a broader range of expertise than is normally the case in transportation and land use planning. For example, retail industry site selection managers would have equal status with regional transportation planners. Most if not all of the significant forces shaping urban form would be considered. It would allow the setting of a planning horizon that reflects the uncertainty inherent in these forces. The land use-transportation scenarios evaluated would not be limited to the regional planning vision and to no-build and build transportation alternatives. Through the iterative process, others would be considered until consensus is reached on a feasible scenario that is compatible with the forces shaping the urban environment. With a multi-disciplinary Delphi panel, broader social equity questions would also likely be considered, as well as a range of opportunity costs.

The process can be open to the public in ways that modeling cannot be. The empirical data, estimates, and assumptions would be available for public inspection. A report might be issued after each step, which would allow stakeholders, including elected officials, the opportunity to provide feedback throughout the effort. Information considered and techniques used would be transferable across regions. A Backcasting Delphi exercise could be executed nationally as well as regionally.


The complex and every-evolving retail structure of the American economy requires that a new method for the evaluation of TOD success be designed and implemented. The method must allow for consideration of the many factors that will determine the transportation and related benefits of compact, mixed-use development. A new method, Backcasting Delphi, which brings together a wide range of expertise and promotes consensus on feasible scenarios, is a possible candidate. It could be a valuable tool in TOD planning and policy-making.


  1. Porter, D. R. Transit-Focused Development: A Synthesis of Research and Experience. TRCP Report No. 20. National Academy Press, 1997.
  2. Project for Public Spaces. The Role of Transit in Creating Livable Metropolitan Communities. TRCP Report No. 22, National Academy Press, 1997.
  3. Niles, J., and D. Nelson. Measuring the Success of Transit-Oriented Development: Retail Market Dynamics and Other Key Determinants. Proceedings of the American Planning Association National Planning Conference, Seattle, Washington, April 24-28, 1999.
  4. Nelson, D., and J. Niles. Market Dynamics and Nonwork Travel Patterns: Obstacles to Transit-Oriented Development? Paper presented at 78th Annual Meeting of the Transportation Research Board, January 1999, Washington, DC (Forthcoming in TRB Transportation Research Record).
  5. Handy, S. Methodologies for Exploring the Link Between Urban Form and Travel Behavior, Transportation Research - D, Vol. 1, No. 2, 1996, pp. 151-165.
  6. Berkeley Planning Journal, The New Urbanism: Expanding the Vision for Design Professions -- A Roundtable Discussion, Berkeley Planning Journal, Vol. 9, 1994, pp. 81-108.
  7. Cervero, R., and P. Bosselmann, Transit Villages: Assessing the Market Potential Through Visual Simulation, Journal of Architectural and Planning Research, Vol. 15, No. 3, Autumn 1998, pp. 181-195.
  8. Boarnet, M. G., and N. S. Compin. Transit-Oriented Development in San Diego County: The Incremental Implementation of a Planning Idea. Journal of the American Planning Association, Vol. 65, No. 1, Winter 1999, pp. 80-94.
  9. Downs, A. New Visions for Metropolitan America. The Brookings Institution and Lincoln Institute of Land Policy, Washington, DC, 1994.
  10. Luscher, D. R. The Odds on TODs: Transit-Oriented Development as a Congestion-Reduction Strategy in the San Francisco Bay Area, Berkeley Planning Journal, Vol. 10, 1995, pp. 55-74.
  11. Fox, T. D., and W. Bowlby. Integrated Decision Support Tool for Evaluating Transit and Land-Use Coordination Policies. Transportation Research Record 1604, 1997, pp. 26-31.
  12. Thompson, G. L., and I. Audirac. TOD’s Importance to Transit: Transit’s Importance to TOD: Planning Scenarios for Sacramento. Washington, DC: 78th Annual Meeting of the Transportation Research Board. 1999
  13. Puget Sound Regional Council. VISION 2020: Growth Strategy and Transportation Plan for the Central Puget Sound Region, Final Environmental Impact Statement, September 1990.
  14. Horowitz, J. L. Travel and Location Behavior: State of the Art and Research Opportunities. Transportation Research - A, Vol. 19A, No. 5/6, 1985, pp. 441-451.
  15. Moore, T, and P. Thorsnes. The Transportation/Land Use Connection: A Framework for Practical Policy, American Planning Association PAS Report 448/449, January 1994.
  16. Hojer, M. Transport Telematics in Urban Systems -- A Backcasting Delphi Study, Transportation Research - D, Vol. 3, No. 6, 1998, pp. 445-464.