I was recently in Copenhagen, where I gave a keynote talk at the UNITE Conference (Uncertainty in Transport Evaluation) at DTU. My thanks to the organizers for inviting me. (I really liked Copenhagen, more on that in a later post).
My talk was titled: Evaluation in a Time of Uncertainty
Abstract: We live in a world where the future is increasingly unpredictable. How should we evaluate transportation investments? Should we even try to evaluate investments in advance? What do we know about what people value? This talk will consider directions in transportation evaluation, and suggestions for better decision-making given uncertainty.
The slide deck is here.
The essential tension in my presentation is the hypothesis that essentially “Long-term forecasting is impossible” vs. my argument that forecasting can only be improved if we “Make forecasters responsible for forecasts.”
Zhang, Lei, and David M. Levinson (2008) Investing for Reliability and Security in Transportation Networks. Transportation Research Record: Journal of the Transportation Research Board #2041 pp.1-10 [doi]
Alternative transportation investment policies can lead to very different network forms in the future. The desirability of a transportation network should be assessed not only by its economic efficiency but also reliability, because the cost of incidental capacity loss in a road network can be massive. This research concerns how investment rules shape the hierarchical structure of roads, and affects network fragility with regard to natural disasters, congestion, and accidents and vulnerability to targeted attacks. A microscopic network growth model predicts the equilibrium road networks under two alternative policy scenarios: investment based on benefit cost analysis or bottleneck removal. A set of Monte-Carlo simulation runs, in which a certain percentage of links are removed according to the type of network degradation analyzed, are carried out to evaluate the equilibrium road networks. It is found that hierarchy exists in road networks for reasons such as economic efficiency, but an overly hierarchical structure has serious reliability problems. Throughout the equilibrating or evolution process, the studied grid network under benefit cost analysis has better efficiency performance, as well as error and attack tolerance. The policy implication from these findings is that benefit-cost analysis should be preferred to myopic bottleneck-removal type of investment rules, no matter how the planning horizon is specified.
Keywords: Transportation network dynamics, road growth model, reliability, vulnerability, fragility, road investment and financing policy