City Rollout Simulator
You are rolling out surge pricing across 20 cities in staggered waves. Design the rollout order, observe noisy revenue data, and estimate the causal impact of your pricing policy.
City Rollout Simulator
This capstone combines experimental design with modern causal inference. In Play mode, you assign cities to treatment waves and try to estimate the true causal effect from observational data. In Design mode, you configure the data-generating process and compare estimation methods: simple DiD, staggered DiD, synthetic control, and event studies.
How to Play
- Assign 20 cities to 4 waves (3 treatment waves + 1 never-treated control)
- Watch revenue data stream in week by week
- Run your DiD estimator on the accumulated data
- See how your rollout design affects estimation bias and precision
Causal Methods
The simulator demonstrates why rollout order matters for causal identification. Cherry-picking high-growth cities for early treatment creates selection bias. Staggered difference-in-differences (following Callaway & Sant’Anna) handles heterogeneous treatment timing correctly, while simple two-period DiD can be severely biased.