By Leslie Vasquez-Guzman
In California, where travel demand models often guide project-level decisions and analysis, a new report finds many are outdated, poorly documented, and ill-suited to the purposes agencies sometimes use them for. Transportation agencies rely on models to forecast traffic and guide billions of dollars in infrastructure investment. Without updates, these tools risk locking in old assumptions about growth and travel behavior rather than helping agencies plan for a more sustainable and efficient future.
Caltrans, the state’s transportation agency, commissioned a review of regional travel demand models to evaluate whether they could support state project work. It found that the models each had base years at least five years old, and most failed to reflect post-pandemic travel habits. Many were not able to account for key effects such as long-term induced travel—new driving that results from expanding road capacity.
The report notes, for example, that most models rely on a single land use scenario, which means they assume a certain base level of transportation investments. “Over the long term,” says the report, “roadway expansion can influence land use patterns by encouraging more spread-out development, leading to longer commutes and greater reliance on personal vehicles.” For a reasonable “no build” comparison, it explains, those new land uses must be removed from the model.
Because of these gaps, project forecasts used for design and environmental review can differ sharply from real-world outcomes once projects are built, and especially if they are not built.
The report also found that limited transparency and poor documentation make many models difficult to evaluate or improve. When users can’t see the underlying assumptions, it’s hard to test a model’s accuracy or understand what’s driving the results. This lack of clarity can lead to decisions based on outdated data rather than current travel behavior or state policy goals.
To address these issues, Caltrans recommends several actions that could help guide modeling in other states. These include updating models with recent data, improving documentation and guidance for agencies, and testing for induced travel and land use changes. The report also calls for a model suitability checklist, better coordination among agencies, and updated standards for project-level modeling to improve consistency and transparency.
The review reinforces a broader lesson: forecasting tools should help agencies understand the outcomes of different choices, not just predict future congestion. As SSTI explored in a recent post on the limits of advanced modeling, sound judgment and clear goals are just as important as modeling accuracy. By aligning models with today’s realities and tomorrow’s priorities, agencies can make better-informed decisions that move transportation systems toward the outcomes they want to achieve.
Photo credit: Esther Gómez on Unsplash. License.