Targeting Mode Shift With an Agent-Based Transport Model
#2 in our "What can you do with an ABM?" series
TLDR: ABMs produce detailed information enabling us to target 'nudge' interventions by finding people with viable alternative modes they could have chosen instead of driving.
The City Modelling Lab and our clients want to encourage people to make more sustainable transport choices — using public transport or active modes (walking or cycling) rather than driving cars. But how do we target our interventions where they might have the biggest impact?
When we run an Agent-Based Model (ABM), each virtual person in our simulation can try hundreds of different ways to travel to their scheduled activities. They search for the travel plan with the combination of mode, route, and timing that works best for them. After each attempt, they keep a note of what they did, what the result was, and remember what worked best.
At the end of the simulation, we have access to every travel plan attempted by every agent and a utility score that tells us how effective each plan was.
We analyse agent choices in their "selected" plans (plans with the highest utility scores) and can compare them to counterfactual decisions the agent could have taken. We can use this analysis to uncover opportunities to change individual behaviour. This analysis is only possible due to the fine-grained level of detail produced by the simulation.
The above graphic from BERTIE shows the home locations of a subset of drivers, specifically those with rail (orange dots) or bus (green dots) trips in their top 10 list of discarded travel plans.
These agents represent drivers with viable public transit options who could be 'nudged' to a more sustainable mode without expensive interventions such as building new infrastructure. Lower-effort interventions like lowering fares or improving access to a local station may be enough to prompt the shift to public transit for these agents.
In this way, agent-based transport models allow us to target the interventions that will produce the biggest bang for the buck.
This is the second post in the series "What can you do with an Agent-Based Model?"
