Integrated Urban Modeling

Dr. Franklin’s early doctoral work considered the mechanisms of analysis that lie behind BCA – in particular, the fact that travel demand modeling usually makes static assumptions about how jobs and households are geographically distributed, even when projecting 20-30 years into the future; this might lead, for example, to an overestimation of travel time savings from a new highway, since many jobs and households might relocate to make use of the new infrastructure. It was exactly such criticism in Salt Lake City, USA, that led the project where he worked, UrbanSim, to be applied and tested there as part of a court settlement between local interest groups favoring and opposing a proposed highway. Estimating the model on historical data, the research team was able to project effects of a variety of land use and transport policies, as documented in their paper, Incorporating Land Use in Metropolitan Transportation Planning.

Transport Equity

Dr. Franklin’s eventual dissertation topic, The Distributional Effects of Congestion Pricing, addressed transport equity, something well-known to be under-addressed by BCA. With the backdrop of a planned bridge replacement in Seattle, USA, where tolls were proposed to help finance construction, Dr. Franklin examined how the potential burdens and benefits would vary across income levels for Seattle area residents. He applied a seldom-used econometric approach that allows “income effects” to be considered in the model of traveler responses to the tolls. The results confirmed past evidence that tolls are income-regressive, but showed that this regressivity is understated when income effects are ignored.

After his dissertation, Dr. Franklin continued pursuing his interest in equity by examining evidence from Stockholm’s cordon-based congestion pricing. This has led to two journal papers and one multi-author book chapter, and a paper in review. In the first article, Behavioral adjustments and equity effects of congestion pricing, and in the book chapter, Traveler Responses to the Stockholm Congestion Pricing Trial, Dr. Franklin’s focus was on comparing the net welfare effects of the congestion pricing across demographic groups. The results indicated that the overall equity effects were less detectable than expected, with no significant difference between men and women and an erratic relationship with income groups.

In the last paper, The Role of Context in the Equity Effects of Congestion Pricing, Dr. Franklin reexamined the evidence using structural equation modeling to account for the mediating effects of a series of contextual factors: a commute that crosses the cordon, flexible work hours, car ownership, and access to a public transport pass. Contrary to the earlier, simpler analysis, this new analysis found significant gender effects in total trip-making in response to the tolls. It also found partial equity effects related to age and income due to car ownership, transit pass, and cordon-crossing, but that these effects worked in opposing directions, resulting in insignificant total equity effects.

Travel Time Reliability

Dr. Franklin’s most recent area of interest is travel time reliability and its value to travelers, as distinct from “expected” travel time. Renewed policy interest in reliability and new theoretical developments have led to a resurgence of research interest, though in practice it is rarely included in BCA. His new work has helped move research toward practice by investigating some of the key obstacles and by identifying what factors are most important when incorporating variability of travel times into BCA.

This research has led to two publications and one paper in review. In the first, Travel Time Reliability for Stockholm Roadways, Franklin and Karlström developed the first statistical model of “mean lateness” – a key variability measure – using road characteristics, showing the importance of the early-peak and late-peak stages of congestion peak hours.

The second paper, Valuations of travel time variability in scheduling versus mean-variance models, with Maria Börjesson and Jonas Eliasson, used a stated-preference study to test a theorized relationship between travelers’ costs of uncertainty versus costs of early and late arrivals. The results showed that they could not, at least in the study’s context – possibly because questions about early and late arrivals might latently suggest that travelers could reschedule their activities en route.

In the final paper, The Reliability Benefits of Congestion Pricing, Dr. Franklin used repeated simulations to estimate the value of reductions in variability associated with Stockholm’s congestion pricing, finding they accounted for up to 40% of all travel time-related cost savings. He also found standard deviation to be a poor proxy for the value of variability, suggesting that much current practice of valuing variability could be flawed.